{"thread":{"id":"15356","subject":"Git User's Survey 2008 partial summary","startedAt":"2008-09-03T14:07:19Z","lastAt":"2008-09-16T17:12:04Z","messageCount":33,"participants":["Jakub Narebski","Shawn O. Pearce","David Brown","H.Merijn Brand","Scott Chacon","Felipe Contreras","David Aguilar","Jan Hudec","Andreas Ericsson","Jonas Fonseca","Anatol Pomozov","Jeff King","david@lang.hm","Nguyen Thai Ngoc Duy","Mark Brown"],"isPatch":false,"patchVersion":null,"patchTotal":null},"messages":[{"id":"89629","messageId":"200809031607.19722.jnareb@gmail.com","threadId":"15356","inReplyTo":null,"subject":"Git User's Survey 2008 partial summary","fromName":"Jakub Narebski","fromEmail":"jnareb@gmail.com","sentAt":"2008-09-03T14:07:19Z","receivedAt":"2008-09-03T14:07:19Z","isPatch":false,"sender":{"key":"jnareb@gmail.com","avatar":"https://avatars.githubusercontent.com/u/2706?v=4"},"body":"This is partial summary of Git User's Survey 2008 after more that 1000 \n(yes, that is more than _thousands_ responses) just after the 3 days\nof running the survey. It is based on \"Analysis\" page for this survey:\n http://www.survs.com/shareResults?survey=M3PIVU72&rndm=OKJQ45LAG8\n\nWe have around 1045 individual responses (including 20 responses in the \n'testing' channel), as compared to 683 individual responses for 2007 \nsurvey, and (I think) 115 answers (Base = 115) for 2006 survey. \nThat is quite a lot, especially after so short time since staring the \nsurvey.\n\n\n03. With which programming languages are you proficient?\n\nAround 939 people answered this question. C is most popular with 59%;\nRuby and surprisingly a bit shell script programming have around second\nplace, with about 52-53%. More people are proficient with Python than \nin Perl by about 1/3-1/4. Very few people (25 responses, around 3%)\nfeel proficient in Tcl/Tk, which means shallow pool of possible git-gui\nand gitk contributors. Java is quite popular with around 42%, which\ngives hope in continued JGit/EGit development.\n\nAnd finally 12 responses (around 1%) of \"I am not a programmer\". That's \nnice. That's very nice.\n\n\n16. I use Git for (check all that apply):\n\nAmong more interesting replies in the \"Other (please specify)\" section\nwere:\n * tracking 3rd party projects, especially with git-svn\n (which probably should be one of possible responses)\n * versioned filesystem\n * distributing teaching examples live\n (which admittedly is form of \"sharing data or sync\" response)\n * real-time collaborative editing (conspire)\n * as a sandbox, to debug changes performed by binary installers\n * storing automated build test results\n\nWork projects and unpaid projects constitute almost the same amount.\n\n\n17. How do you obtain Git?\n\nMost people use binary packages (59%), with source packages and tarballs\nsecond (28%) and pull from main repository third (22%).\n\nNote that in this version of survey this question is multiple choice \n(multiple answers), as opposed to single choice in earlier surveys.\n\n\n18. What operating system do you use Git on?\n\nLinux, then MacOS X, then MS Windows. In \"Other, please specify\"\nthere are two responses of 'iPhone OS'... can one use git on iPhone?\n\n\n21. Which porcelains / interfaces / implementations do you use?\n\nMost people (99%, 862/869) use core-git in some form, perhaps in \naddition to some other interface. Cogito no longer has large\namount of users: only 5; it looks like people noticed finally that\nis deprecated and not maintained any longer. Among patch management \ninterfaces StGIT dominates with 24 (3% share), then newcomer TopGit \nwith 13, and last Guilt with 6; but the last part can be caused by \npeople not noticing it among answers, as I saw a few of \"Guilt\"/\"guilt\"\namong free-form \"Other (please specify)\". Patchy GIT (pg), long\ndeprecated, has happily no users.\n\n\"My own scrips\" with 117 / 13% responses shows very well either good\nGit scriptability, or lack of good UI ;-). I guess it is the former.\n\n\"Other\" responses include Grit, which I forgot is at least partially \n(re)implementation of Git, git-wtf and gc-utils... but some of 'other' \nresponses include tools which are (I think) neither porcelain, like \ngitosis or gibak, or git-cola, or magit, but more of a GUI.\n\nPeople who do not use \"git (core)\" implementation use StGIT, own \nscripts, or Easy Git.\n\n\n22. Which git GUI (commit tool or history viewer, or both) do you use?\n\nI think it should have been stated more clearly that this list includes \nalso integration with IDE, RAD and editors. Here among tools which I \nforgot to add are magit which is another Emacs interface, and Textmate \nbundle.\n\nHmmm... only 75% people use command line; it means that graphical tools \nimproved. Tools which are distributed along git, namely Tcl/Tk based \ngit-gui and gitk dominates. Bit suprising for me is third place (if we \nexclude CLI) of GitNub, and 6% of tig users. QGit is fourth with 11%;\nGiggle has more users (26) than GitView (4). Quite new GitX has \nsurprisingly as much as 16 (2%) users. \n\nAmong 'others' there is nice and a bit curious \"testing the integration \nin upcoming KDevelop 4\" (which was GSoC 2008 project for KDE).\n\n\n23. Which git web interface do you use for your projects/have installed?\n\nIt was meant to be about web interface _you_ run, and not only use as a \npart of third party git hosting site you use. It is a bit strange that \npeople put GitHub there, unless GitHub source code was made available, \nor is it possible to deploy/install clone of GitHub yourself.\n\nAs before gitweb dominates with 80% of share, then Gitorious with 14% \nand next cgit with 11% (30 responses). \n\nAmong 'others' there are Trac (I guess Trac plugin), Redmine \n(I guess also plugin), ikiwiki and \"my own, not public yet\".\n\n\n\n50. Would commercial (paid) Git support from a support vendor be of \ninterest to you/your organization?\n\n746 individual responses, 9% yes (66 replies), 69% no, 22% not \napplicable.\n\n\nTo be continued...\n-- \nJakub Narebski\nPoland\n"},{"id":"89631","messageId":"20080903144552.GA27682@spearce.org","threadId":"15356","inReplyTo":"200809031607.19722.jnareb@gmail.com","subject":"Re: Git User's Survey 2008 partial summary","fromName":"Shawn O. Pearce","fromEmail":"spearce@spearce.org","sentAt":"2008-09-03T14:45:52Z","receivedAt":"2008-09-03T14:45:52Z","isPatch":false,"sender":{"key":"spearce@spearce.org","avatar":"https://avatars.githubusercontent.com/u/34844?v=4"},"body":"Jakub Narebski wrote:\n> This is partial summary of Git User's Survey 2008 [...]\n> \n> 03. With which programming languages are you proficient?\n> \n> Around 939 people answered this question. C is most popular with 59%;\n> Ruby and surprisingly a bit shell script programming have around second\n> place, with about 52-53%. More people are proficient with Python than \n> in Perl by about 1/3-1/4. Very few people (25 responses, around 3%)\n> feel proficient in Tcl/Tk, which means shallow pool of possible git-gui\n> and gitk contributors. Java is quite popular with around 42%, which\n> gives hope in continued JGit/EGit development.\n\nYea, I liked seeing that result from our user base. 42% of the\nrespondants know Java as a language. Not all of them may like\nprogramming in it for fun, but at least they claim they know it. :-)\n\nIn the IDE part Eclipse received 13% (124 responses), which is\nalmost close to Emacs (20%, 193 responses). Sadly only 2% (19\nresponses) use EGit. There's probably a lot of reasons for that.\nPossibly the \"egit_sux\" tag on repo.or.cz. Or our lack of a stable\nupdate site that users can install reasonably stable versions from.\n\n\nI think we've picked up 3 new contributors recently to the JGit/EGit\nproject. Jonas Fonseca has started to use JGit inside of a NetBeans\nIDE plugin. Today he posted 2 patches, hopefully the first of many.\n\nTor Arne Vestb� has been working on improving EGit's resource\ndecorator. His fork on repo.or.cz is a few days old, but it has\nsome really good potential in there. I'm looking forward to seeing\nthe final series from his work.\n\nCharles O'Farrell joined us a few weeks back and added support for\nref deletion, along with CLI for \"jgit branch -d/-D\". I think his\ntime is rather limited, but the contributions are most appreciated.\n\nAnd of course Marek Zawirski is still contributing, but I think\nhe's back in classes so his time is a little bit more constrained.\n\nMonths back I think someone asked about including JGit in a\ncommerically licensed product. Since its 3-clause BSD its quite\npossible to do. But I haven't seen anything materialize yet on\nthat front.\n\n-- \nShawn.\n"},{"id":"89632","messageId":"20080903150018.GA23061@linode.davidb.org","threadId":"15356","inReplyTo":"200809031607.19722.jnareb@gmail.com","subject":"Re: Git User's Survey 2008 partial summary","fromName":"David Brown","fromEmail":"git@davidb.org","sentAt":"2008-09-03T15:00:19Z","receivedAt":"2008-09-03T15:00:19Z","isPatch":false,"sender":{"key":"git@davidb.org","avatar":"https://gravatar.com/avatar/94c86a2938470a74c2eac5e2b69afc0871f79a660295c02219597aba8cb101c1?d=mp&s=160"},"body":"On Wed, Sep 03, 2008 at 04:07:19PM +0200, Jakub Narebski wrote:\n\n>18. What operating system do you use Git on?\n>\n>Linux, then MacOS X, then MS Windows. In \"Other, please specify\"\n>there are two responses of 'iPhone OS'... can one use git on iPhone?\n\nProbably not too difficult if the phone has been hacked, since it's\nmostly a stripped down OSX.\n\nDavid\n"},{"id":"89635","messageId":"20080903172050.61d510d6@pc09.procura.nl","threadId":"15356","inReplyTo":"20080903144552.GA27682@spearce.org","subject":"Re: Git User's Survey 2008 partial summary","fromName":"H.Merijn Brand","fromEmail":"h.m.brand@xs4all.nl","sentAt":"2008-09-03T15:20:50Z","receivedAt":"2008-09-03T15:20:50Z","isPatch":false,"sender":{"key":"h.m.brand@xs4all.nl","avatar":"https://gravatar.com/avatar/5b8f83ee35c427a646cbea3b104346e00ab3663b99bbf435cddeb75cd4b3857b?d=mp&s=160"},"body":"On Wed, 3 Sep 2008 07:45:52 -0700, \"Shawn O. Pearce\"\n wrote:\n\n> Jakub Narebski wrote:\n> > This is partial summary of Git User's Survey 2008 [...]\n> > \n> > 03. With which programming languages are you proficient?\n> > \n> > Around 939 people answered this question. C is most popular with 59%;\n> > Ruby and surprisingly a bit shell script programming have around second\n> > place, with about 52-53%. More people are proficient with Python than \n> > in Perl by about 1/3-1/4. Very few people (25 responses, around 3%)\n> > feel proficient in Tcl/Tk, which means shallow pool of possible git-gui\n> > and gitk contributors.\n\nRewrite them in perl or python and get more patches?\nThe fact that it is not perl withheld me from finding solutions to\nproblems I still have with the git-gui gitk combination\n\n> > Java is quite popular with around 42%, which\n> > gives hope in continued JGit/EGit development.\n\nI would love to use git in Eclipse, but I tried again this week and\nfound the following problems\n\n1. egit web site does not give you any clue about installation or\n integration with eclipse or how that could be done at all\n2. I found no suitable eclipse-egit package for OpenSUSE 10.3, and the\n packages I /did/ find depended on packages not available for SuSE\n 10.3\n\n> Yea, I liked seeing that result from our user base. 42% of the\n> respondants know Java as a language. Not all of them may like\n> programming in it for fun, but at least they claim they know it. :-)\n\nI *hate* programming in Eclipse, but some java projects here require me\nto work with it. My preferred IDE is elvis (a vi clone)\n\n> In the IDE part Eclipse received 13% (124 responses), which is\n> almost close to Emacs (20%, 193 responses). Sadly only 2% (19\n> responses) use EGit. There's probably a lot of reasons for that.\n> Possibly the \"egit_sux\" tag on repo.or.cz. Or our lack of a stable\n> update site that users can install reasonably stable versions from.\n\n/me fiercely nods!\n\n-- \nH.Merijn Brand Amsterdam Perl Mongers http://amsterdam.pm.org/\nusing & porting perl 5.6.2, 5.8.x, 5.10.x, 5.11.x on HP-UX 10.20, 11.00,\n11.11, 11.23, and 11.31, SuSE 10.1, 10.2, and 10.3, AIX 5.2, and Cygwin.\nhttp://mirrors.develooper.com/hpux/ http://www.test-smoke.org/\nhttp://qa.perl.org http://www.goldmark.org/jeff/stupid-disclaimers/\n"},{"id":"89637","messageId":"d411cc4a0809030841n7defd31frbbab2a1b7ab824d7@mail.gmail.com","threadId":"15356","inReplyTo":"200809031607.19722.jnareb@gmail.com","subject":"Re: Git User's Survey 2008 partial summary","fromName":"Scott Chacon","fromEmail":"schacon@gmail.com","sentAt":"2008-09-03T15:41:56Z","receivedAt":"2008-09-03T15:41:56Z","isPatch":false,"sender":{"key":"schacon@gmail.com","avatar":"https://gravatar.com/avatar/9b13a8a078e1dcf8588c4eea9554445d51ebed6c41b51f56f4d96738130b05c6?d=mp&s=160"},"body":"On Wed, Sep 3, 2008 at 7:07 AM, Jakub Narebski wrote:\n> This is partial summary of Git User's Survey 2008 after more that 1000\n> (yes, that is more than _thousands_ responses) just after the 3 days\n> of running the survey. It is based on \"Analysis\" page for this survey:\n> http://www.survs.com/shareResults?survey=M3PIVU72&rndm=OKJQ45LAG8\n>\n\nIs there a way to download the raw data? There may be some\ninteresting correlations between answers in different fields I would\nlike to look for in the data, but it's too aggregated here. I heard\nit would be available at the end, but can I get it from somewhere now\nas well?\n\nScott\n"},{"id":"89640","messageId":"200809031800.27739.jnareb@gmail.com","threadId":"15356","inReplyTo":"d411cc4a0809030841n7defd31frbbab2a1b7ab824d7@mail.gmail.com","subject":"Re: Git User's Survey 2008 partial summary","fromName":"Jakub Narebski","fromEmail":"jnareb@gmail.com","sentAt":"2008-09-03T16:00:26Z","receivedAt":"2008-09-03T16:00:26Z","isPatch":false,"sender":{"key":"jnareb@gmail.com","avatar":"https://avatars.githubusercontent.com/u/2706?v=4"},"body":"On Wed, 3 September 2008, Scott Chacon wrote:\n> On Wed, Sep 3, 2008 at 7:07 AM, Jakub Narebski wrote:\n> >\n> > This is partial summary of Git User's Survey 2008 after more that 1000\n> > (yes, that is more than _thousands_ responses) just after the 3 days\n> > of running the survey. It is based on \"Analysis\" page for this survey:\n> > http://www.survs.com/shareResults?survey=M3PIVU72&rndm=OKJQ45LAG8\n> \n> Is there a way to download the raw data? There may be some\n> interesting correlations between answers in different fields I would\n> like to look for in the data, but it's too aggregated here. I heard\n> it would be available at the end, but can I get it from somewhere now\n> as well?\n\nI, as survey administrator, can export survey data to CSV or Excel\nformats, but currently I cannot export URL to such data. So you can\neither ask me to send you such data[*1*] (CSV for around 700 individual\nresponses weights 1.7MB, and 272KB after gzipping), or sign in\nto Survs.com (which might, or might not, require invite) and ask me\nto add to \"git\" role/account/group.\n\nOn the other hand you can use different filters (again, I think only\nsurvey administrator can do this, or whatever role is it) to limit\nsummarized replies, for example show summary only for users which\nuse Git on MS Windows (answered \"What operating system...\" question,\nand the answer was \"MS Windows (Cygwin)\" or \"MS Windows (msysGit)\".\n\nI can also enable individual responses in shared view (above link),\nso you can see reply by reply; but I don't think this is particularly\nuseful, especially with 1000+ responses (21 in 'testing' channel).\n\n\n[*1*] I will not send this to git mailing list...\n-- \nJakub Narebski\nPoland\n"},{"id":"89641","messageId":"94a0d4530809030925t33d4260bof169372d65717af4@mail.gmail.com","threadId":"15356","inReplyTo":"20080903172050.61d510d6@pc09.procura.nl","subject":"Re: Git User's Survey 2008 partial summary","fromName":"Felipe Contreras","fromEmail":"felipe.contreras@gmail.com","sentAt":"2008-09-03T16:25:17Z","receivedAt":"2008-09-03T16:25:17Z","isPatch":false,"sender":{"key":"felipe.contreras@gmail.com","avatar":"https://avatars.githubusercontent.com/u/8358?v=4"},"body":"On Wed, Sep 3, 2008 at 6:20 PM, H.Merijn Brand wrote:\n> On Wed, 3 Sep 2008 07:45:52 -0700, \"Shawn O. Pearce\"\n> wrote:\n>\n>> Jakub Narebski wrote:\n>> > This is partial summary of Git User's Survey 2008 [...]\n>> >\n>> > 03. With which programming languages are you proficient?\n>> >\n>> > Around 939 people answered this question. C is most popular with 59%;\n>> > Ruby and surprisingly a bit shell script programming have around second\n>> > place, with about 52-53%. More people are proficient with Python than\n>> > in Perl by about 1/3-1/4. Very few people (25 responses, around 3%)\n>> > feel proficient in Tcl/Tk, which means shallow pool of possible git-gui\n>> > and gitk contributors.\n>\n> Rewrite them in perl or python and get more patches?\n> The fact that it is not perl withheld me from finding solutions to\n> problems I still have with the git-gui gitk combination\n\nRuby!\n\n-- \nFelipe Contreras\n"},{"id":"89713","messageId":"402731c90809031943n181ad6fbw366f90e80aaca2a@mail.gmail.com","threadId":"15356","inReplyTo":"94a0d4530809030925t33d4260bof169372d65717af4@mail.gmail.com","subject":"Re: Git User's Survey 2008 partial summary","fromName":"David Aguilar","fromEmail":"davvid@gmail.com","sentAt":"2008-09-04T02:43:20Z","receivedAt":"2008-09-04T02:43:20Z","isPatch":false,"sender":{"key":"davvid@gmail.com","avatar":"https://avatars.githubusercontent.com/u/13196?v=4"},"body":"On Wed, Sep 3, 2008 at 9:25 AM, Felipe Contreras\n wrote:\n> On Wed, Sep 3, 2008 at 6:20 PM, H.Merijn Brand wrote:\n>> On Wed, 3 Sep 2008 07:45:52 -0700, \"Shawn O. Pearce\"\n>> wrote:\n>>\n>>> Jakub Narebski wrote:\n>>> > This is partial summary of Git User's Survey 2008 [...]\n>>> >\n>>> > 03. With which programming languages are you proficient?\n>>> >\n>>> > Around 939 people answered this question. C is most popular with 59%;\n>>> > Ruby and surprisingly a bit shell script programming have around second\n>>> > place, with about 52-53%. More people are proficient with Python than\n>>> > in Perl by about 1/3-1/4. Very few people (25 responses, around 3%)\n>>> > feel proficient in Tcl/Tk, which means shallow pool of possible git-gui\n>>> > and gitk contributors.\n>>\n>> Rewrite them in perl or python and get more patches?\n>> The fact that it is not perl withheld me from finding solutions to\n>> problems I still have with the git-gui gitk combination\n>\n> Ruby!\n>\n> --\n> Felipe Contreras\n\nThere's already a python git-gui:\n http://cola.tuxfamily.org/\n\nPyQt is a very mature library, which is one of the primary reasons I\nchose Python.\nDoes Ruby have any good and mature UI libraries? I know it's all the\nrage for web stuff, but I haven't heard too much about people using it\nfor GUIs.\n\n-- \n David\n"},{"id":"89763","messageId":"200809041523.40601.jnareb@gmail.com","threadId":"15356","inReplyTo":"200809031607.19722.jnareb@gmail.com","subject":"Re: Git User's Survey 2008 partial summary, part 2","fromName":"Jakub Narebski","fromEmail":"jnareb@gmail.com","sentAt":"2008-09-04T13:23:37Z","receivedAt":"2008-09-04T13:23:37Z","isPatch":false,"sender":{"key":"jnareb@gmail.com","avatar":"https://avatars.githubusercontent.com/u/2706?v=4"},"body":"This is partial summary of Git User's Survey 2008 after more that 1600 \n(yes, that is more 1.6 _thousands_ responses) just after the 4 days\nof running the survey. It is based on \"Analysis\" page for this survey:\n http://www.survs.com/shareResults?survey=M3PIVU72&rndm=OKJQ45LAG8\n \nWe have around 1606 individual responses (excluding 21 responses in the \n'testing' channel), as compared to 683 individual responses for 2007 \nsurvey, and (I think) 115 answers (Base = 115) for 2006 survey. \nThat is a lot, especially after so short time since staring the \nsurvey.\n\n\n03. With which programming languages are you proficient?\n\nSituation changed a bit since last announcement. Nor Ruby with 59%\ndominates a bit over C with 55%. I guess this is caused by GitHub\ncrowd: announcement about this survey was posted on GitHub blog\n(thanks defunkt!).\n \nThere are now 22 responses (around 1%) of \"I am not a programmer\".\nThis means that you don't need to be a programmer to use Git.\n\nPopular (from what I have glanced) languages in the 'other' section\ninclude Erlang, Lua, OCaml, Haskell, Objective-C; there is Scala, Vala, \nSmalltalk, R, ActionScript, Prolog, SQL, Forth, and even TeX/LaTeX \nmentioned; someone even wrote Brainfuck here. \n\n\n04. How did you hear about Git?\n\nHere 'blog entry' and 'some project uses it' with 38% dominates over \n'Linux kernel news' and 'word of mouth' with 22-24%.\n\nAmong 'other' there is quite popular Google Tech Talk about Git by Linus \nTorvalds, available on YouTube (which admittedly probably should be as \none of choices), LKML (Linix kernel mailing list) which probably some \nput as 'Linux kernel news' and some not, FLOSS podcast, GitHub, Ruby \nadopting Git. Among more unique and interesting 'other' replies there \nis \"Faculty Advisor at NSF REU\" and \"school Professor\", \"Ruby community \njust went apeshit all of a sudden\" ;-) and \"Ruby On Rails switching to \ngit overnight\"; some individual people line 'Randall Schwartz' or \n'Scott Chacon'.\n\n\n08. Which Git version(s) are you using? (multiple choices)\n\nMost people use current version of git: 70% use git 1.5.x, and 40% use \ngit 1.6.x, and only 3 people use pre 1.3.x version. Note that some\npeople use different versions on different machines, or on different \noperating systems, so the numbers do not add to 100%.\n\nAround 8% (123 replies) use 'master' branch. Sorry Junio, it looks like \nit wouldn't be easy to find/detect bugs as soon as they hit 'master', \nbefore releasing. (Unfortunately this survey lack question to find how \nmany people run *-rcN (release candidate) version, but it is very long \nas it is with 60 questions and 55 minutes average time of answering \nquestion, and anegdotical evidence (comments) of requiring around 15 \nminutes to fill it without spending much time on free-form questions).\n\n\n25. How do you publish/propagate your changes? (multiple choices)\n\nMost people (92%) simply push to publishing public repository. Second \nis 'pull request' with 32%; I guess this includes GitHub social thingy.\ngit-svn as form of publishing changes is third with 27%; thanks Eric \nWong! Format-patch + email is only fourth with 22%, but quite close to \ngit-svn and pull-request.\n\nFor publishing to other SCMs (32 replies, 2% or responses) people use \ngit-bzr, git-p4, git-cvsexportcommit, git-acu (AccuRev); some publish \nto Mercurial, ClearCase or Visual SourceSafe.\n\n\nTo be continued...\n\nP.S. Which non free-form questions are you interested most, BTW?\n-- \nJakub Narebski\nPoland\n"},{"id":"89876","messageId":"20080905221731.GI15520@efreet.light.src","threadId":"15356","inReplyTo":"402731c90809031943n181ad6fbw366f90e80aaca2a@mail.gmail.com","subject":"Re: Git User's Survey 2008 partial summary","fromName":"Jan Hudec","fromEmail":"bulb@ucw.cz","sentAt":"2008-09-05T22:17:31Z","receivedAt":"2008-09-05T22:17:31Z","isPatch":false,"sender":{"key":"bulb@ucw.cz","avatar":null},"body":"On Wed, Sep 03, 2008 at 19:43:20 -0700, David Aguilar wrote:\n> On Wed, Sep 3, 2008 at 9:25 AM, Felipe Contreras\n> wrote:\n> > On Wed, Sep 3, 2008 at 6:20 PM, H.Merijn Brand wrote:\n> >> On Wed, 3 Sep 2008 07:45:52 -0700, \"Shawn O. Pearce\"\n> >> wrote:\n> >>\n> >>> Jakub Narebski wrote:\n> >>> > This is partial summary of Git User's Survey 2008 [...]\n> >>> >\n> >>> > 03. With which programming languages are you proficient?\n> >>> >\n> >>> > Around 939 people answered this question. C is most popular with 59%;\n> >>> > Ruby and surprisingly a bit shell script programming have around second\n> >>> > place, with about 52-53%. More people are proficient with Python than\n> >>> > in Perl by about 1/3-1/4. Very few people (25 responses, around 3%)\n> >>> > feel proficient in Tcl/Tk, which means shallow pool of possible git-gui\n> >>> > and gitk contributors.\n> >>\n> >> Rewrite them in perl or python and get more patches?\n> >> The fact that it is not perl withheld me from finding solutions to\n> >> problems I still have with the git-gui gitk combination\n> >\n> > Ruby!\n> >\n> > --\n> > Felipe Contreras\n> \n> There's already a python git-gui:\n> http://cola.tuxfamily.org/\n> \n> PyQt is a very mature library, which is one of the primary reasons I\n> chose Python.\n\nSorry, but I disagree. Tried PyQt, been hugely disapointed. Boils down to any\nthing that can make Python (or, for that matter, any) interpreter segfault\nbeing totally broken.\n\nBut as far as Qt goes, I would really just stick with C++. Python or Ruby\nhave some advantage, but I am not sure it's that big to offset the fact, that\na lot of code already exists in QGit.\n\n> Does Ruby have any good and mature UI libraries? I know it's all the\n> rage for web stuff, but I haven't heard too much about people using it\n> for GUIs.\n\nQt? I believe Ruby Qt bindings are in better shape (properly handle Qt\ndeleting objects under Ruby's hands).\n\n-- \n\t\t\t\t\t\t Jan 'Bulb' Hudec \n"},{"id":"89888","messageId":"200809060422.34779.jnareb@gmail.com","threadId":"15356","inReplyTo":"200809031607.19722.jnareb@gmail.com","subject":"Git User's Survey 2008 partial summary, part 3","fromName":"Jakub Narebski","fromEmail":"jnareb@gmail.com","sentAt":"2008-09-06T02:22:32Z","receivedAt":"2008-09-06T02:22:32Z","isPatch":false,"sender":{"key":"jnareb@gmail.com","avatar":"https://avatars.githubusercontent.com/u/2706?v=4"},"body":"This is partial summary of Git User's Survey 2008 after more that 2000 \n(yes, that is more than _two thousands_ responses) just after the 5 days\nof running the survey. It is based on \"Analysis\" page for this survey:\n http://www.survs.com/shareResults?survey=M3PIVU72&rndm=OKJQ45LAG8\n\nWe have around 2040 individual responses (excluding 21 responses in the \n'testing' channel), as compared to 683 individual responses for 2007 \nsurvey, and (I think) 115 answers (Base = 115) for 2006 survey. \nThat is quite a lot, especially after only five days (or so) since \nstaring the survey.\n \n\nGetting started with Git:\n~~~~~~~~~~~~~~~~~~~~~~~~~\n\n05. Did you find Git easy to learn?\n\nReplies look like nice gaussian-like distribution, centered around the \nmiddle, but with slight bias towards 'easy' part, with 16% hard vs 24% \neasy, and 2% very hard vs 6% very easy.\n\nThe 2% (33) for 'very hard' can be a bit of a concern...\n\n\n09. How long do you use Git?\n\nMajority of people use Git for about from 3 months to 1 year (29% for \n3-6 months, 26% for 6-12 months). The graph is centered around this \nvalues, too. Few people (58, 3%) said that they use Git more than two \nyears, fewer still (41, 2%) stated that they use it from the very \nbeginning.\n\nThis hopefully means that most people had contact with git after it had \nimproved its UI.\n\n\n10. Rate your own proficiency with Git\n\nHere again the distribution is almost perfectly symmetric. Most people \nsay that they are proficient with Git enough for everyday use (44%), \nslightly more (23%) can offer advice than selected 'casual, needs \nadvice', while slightly more (7%) thinks themself novices than people \nwho think that know Git very well (4%).\n\nFor this survey it seems to imply that results would not be biased \n(scewed) by people who know Git well to very well.\n\n\n11. Does git.git repository include code produced by you?\n\n154 says yes, 1674 says no, 227 skipped this question.\n\nIt is not just git developers that answer this question... which is kind \nof obvious because we do not have that many developers (I wish).\n\n\nHow do you use Git:\n~~~~~~~~~~~~~~~~~~~\n\n18. What operating system do you use Git on? (multiple answers)\n\nThis question was made multiple choices (one can use Git on different \noperating systems; either multi-boot, or multiple machines), as \ncompared to free-form question in 2006 and 2007 survey.\n\nLinux (or should we say GNU/Linux?) dominates with 82%, second is\nMacOS X (Darwin) with 59%, and MS Windows is only third with 25%\n(here msysGit is more popular with 15% than Cygwin with 10%).\nIf we consider fact that MS Windows dominates desktop, one can see that \nGit is not very popular choice for Windows machines, probably because \nof perceived and real shortcomings when trying to run it on non-POSIX \noperating system. The domination of Linux is not strange considering \nfact that Git was created for the purpose of being distributed version \ncontrol system for Linux kernel, and is designed around Linux features \n(fast fork, fast stat, prototyping using scripts, etc.).\n\nAmong 'other' operating systems there is iPhone, iPod, \nMS Windows/AndLinux,... but also for example Ubuntu or Debian Sarge,\nwhich clearly should be marked as Linux instead.\n\n\n19. What hardware platforms do you use Git on? (multiple answers)\n\nHere slightly more people (88%) use 32-bit machines than 64-bit machines \n(61%)... if we can say that, as it is multiple choice question. Few \npeople chose 'portable/handheld' (51, 3%), fewer still selected \n'other' (8 replies). \n\nHere 'other' I think mean: I don't know which option to chose. There is \none OLPC-XO, but also PowerPC and PPC/G4.\n\n\n20. What editor, IDE or RAD you use working with Git?\n\nCurrently Vim, with 48% or 841 responses, wins by narrow margin over \nTextMate which has 44% (782 responses). Curious.\n\nEmacs is third with 18%, followed closely by Eclipse with 13%. More \nthan 50 replies (responses) also have NetBeans IDE, XCode, MS Visual \nStudio and Notepad++.\n\nAmong 'other (please specify)' we have nano, Kate, Gedit, Geany, \ne-Editor (aka. TextMate for Windows) or just e, jEdit, SlickEdit, \nSharpDevelop, SciTE, Code::Blocks, joe, ActiveState Komodo, Aptana and \na few others; even one 'ed' ;-)\n\nNote that quite a few people wrote about integration with IDE/editor in \nthe \"35. What tools (or kind of tools) would you like to see Git \nsupport in?\" question.\n\nTo be continued...\n\n-- \nJakub Narebski\nPoland\n"},{"id":"89889","messageId":"20080906041733.GA18930@gmail.com","threadId":"15356","inReplyTo":"20080905221731.GI15520@efreet.light.src","subject":"Re: Git User's Survey 2008 partial summary","fromName":"David Aguilar","fromEmail":"davvid@gmail.com","sentAt":"2008-09-06T04:17:34Z","receivedAt":"2008-09-06T04:17:34Z","isPatch":false,"sender":{"key":"davvid@gmail.com","avatar":"https://avatars.githubusercontent.com/u/13196?v=4"},"body":"On 0, Jan Hudec wrote:\n> > \n> > There's already a python git-gui:\n> > http://cola.tuxfamily.org/\n> > \n> > PyQt is a very mature library, which is one of the primary reasons I\n> > chose Python.\n> \n> Sorry, but I disagree. Tried PyQt, been hugely disapointed. Boils down to any\n> thing that can make Python (or, for that matter, any) interpreter segfault\n> being totally broken.\n\n\nThat's curious, I've never seen a single segfault.\nYou don't happen to run suse or fedora, do you?\nI've not experienced your pains once on debian.\n\n\n> But as far as Qt goes, I would really just stick with C++. Python or Ruby\n> have some advantage, but I am not sure it's that big to offset the fact, that\n> a lot of code already exists in QGit.\n\nI have no comment here -- for me it boils down to the fact\nthat I end up writing much less code in Python vs. C++. GUIs\nare not performance critical. Most of the time is spent in\nthe event loop waiting for user input. Ditto for Ruby, tcltk,\netc --the benefit is in writing less code that does more. I\nsure as hell wouldn't write something performance critical in\nthose languages, but for something simple like a GUI it makes\nperfect sense.\n\ngit-cola and qgit actually do different things. qgit has\nan awesome history viewer. git-cola is primarily for doing\nstuff like splitting apart changes into separate commits, etc.\nI tend to work things up into a working state and then commit\nthings separately. For example, there might be one commit\nthat does some refactoring and a second commit that uses the\nrefactored code, etc. git-cola started out with emulating\ngit-gui's featureset though it has surpassed it now, which is\nwhy its focus is on the commit workflow.\n\n\n> > Does Ruby have any good and mature UI libraries? I know it's all the\n> > rage for web stuff, but I haven't heard too much about people using it\n> > for GUIs.\n> \n> Qt? I believe Ruby Qt bindings are in better shape (properly handle Qt\n> deleting objects under Ruby's hands).\n\nRight -- if you got segfaults due to object deletion it's\nbecause you let things go out of scope. I do not doubt that\nthe ruby bindings handle this better, but I have not ever\nseen the problems you describe so I wouldn't know.\nRubyQt looks interesting but the differences are superficial\ngiven that I've not run into any problems. As far as suckage\ngoes, ruby and python both suck equally so the choice is\nimmaterial ;)\n\nAnyways, this is getting off topic for the git list so I'll\nstop here. If you have some pyqt examples that demonstrate\nsegfaults then I'm sure phil (pyqt author) would be happy\nto see them, but those details are probably left off-list or\non the pyqt devel list. \n\nciao,\n\n-- \n\n\tDavid\n"},{"id":"89892","messageId":"20080906051530.GB28035@spearce.org","threadId":"15356","inReplyTo":"200809060422.34779.jnareb@gmail.com","subject":"Re: Git User's Survey 2008 partial summary, part 3","fromName":"Shawn O. Pearce","fromEmail":"spearce@spearce.org","sentAt":"2008-09-06T05:15:30Z","receivedAt":"2008-09-06T05:15:30Z","isPatch":false,"sender":{"key":"spearce@spearce.org","avatar":"https://avatars.githubusercontent.com/u/34844?v=4"},"body":"Jakub Narebski wrote:\n> \n> Note that quite a few people wrote about integration with IDE/editor in \n> the \"35. What tools (or kind of tools) would you like to see Git \n> support in?\" question.\n\nBoth Eclipse and NetBeans are getting on board with JGit and trying\nto make viable plugins from it. So hopefully by the next survey\nthis particular question will have been answered for a decent\nprecentage of our IDE toting users.\n\nActually, I'm rather excited by what's going on in the Git IDE\nspace these days. The NetBeans plugin is seeing more activity,\nand since GSoC so has EGit. Both projects have been steadily\npicking up contributors. Quite a good thing.\n\n-- \nShawn.\n"},{"id":"89897","messageId":"48C23EE0.8090507@op5.se","threadId":"15356","inReplyTo":"20080906051530.GB28035@spearce.org","subject":"Re: Git User's Survey 2008 partial summary, part 3","fromName":"Andreas Ericsson","fromEmail":"ae@op5.se","sentAt":"2008-09-06T08:27:12Z","receivedAt":"2008-09-06T08:27:12Z","isPatch":false,"sender":{"key":"ae@op5.se","avatar":"https://gravatar.com/avatar/426e89595c75a8f5252dd0c989e5fabe5bcac616e68557427ad9aef6b0ca342a?d=mp&s=160"},"body":"Shawn O. Pearce wrote:\n> Jakub Narebski wrote:\n>> Note that quite a few people wrote about integration with IDE/editor in \n>> the \"35. What tools (or kind of tools) would you like to see Git \n>> support in?\" question.\n> \n> Both Eclipse and NetBeans are getting on board with JGit and trying\n> to make viable plugins from it. So hopefully by the next survey\n> this particular question will have been answered for a decent\n> precentage of our IDE toting users.\n> \n> Actually, I'm rather excited by what's going on in the Git IDE\n> space these days. The NetBeans plugin is seeing more activity,\n> and since GSoC so has EGit. Both projects have been steadily\n> picking up contributors. Quite a good thing.\n> \n\nThe geany ide has some limited git support as well. Since it's\nwritten in C, I believe a libified git would boost its level of\nsupport tremendously.\n\n-- \nAndreas Ericsson andreas.ericsson@op5.se\nOP5 AB www.op5.se\nTel: +46 8-230225 Fax: +46 8-230231\n"},{"id":"89929","messageId":"200809070017.09463.jnareb@gmail.com","threadId":"15356","inReplyTo":"200809031607.19722.jnareb@gmail.com","subject":"Git User's Survey 2008 partial summary, part 4 - how do we use Git","fromName":"Jakub Narebski","fromEmail":"jnareb@gmail.com","sentAt":"2008-09-06T22:17:08Z","receivedAt":"2008-09-06T22:17:08Z","isPatch":false,"sender":{"key":"jnareb@gmail.com","avatar":"https://avatars.githubusercontent.com/u/2706?v=4"},"body":"This is partial summary of Git User's Survey 2008 after more that 2000 \n(yes, that is more than _two thousands_ responses) just after the 6 days\nof running the survey. It is based on \"Analysis\" page for this survey:\n http://www.survs.com/shareResults?survey=M3PIVU72&rndm=OKJQ45LAG8\n\nPlease think of ways to examine this data, and about conclusions that we \ncan draw from those statistics.\n\n\n\n27 & 28. How often do you use the following forms of git commands or\n extra git tools? (Matrix - One answer per row)\n + never / rarely / sometimes / often\n\nFirst, even though those questions looks a bit terryfying on first\nglance, with this large matrix of commands, most people (around 3/4)\nfilled those two questions: 1562 responses with 540 skipped for 27.,\nand 1454 total and 648 skipped for its continuation, question 28.\n\nThis large matrix had the additional advantage of teaching newcomers\nto git about some rarer or newer git commands or versions of git\ncommands.\n\nBelow there is very simple statistics based on individual columns,\nwithout taking into consideration any correlation between columns, or\nany relation. For example it doesn't take into account that for some\n'often' is somene's else 'sometimes'.\n\n\nTop 10 often used commands\n\n===================================\nCommand + options | often [%]\n-----------------------------------\ngit push | 72% (1120)\ngit add + git commit | 65% (1012)\ngit commit -a | 63% (981)\ngit pull (no remote) | 44% (682)\ngit pull | 37% (578)\ngit commit ... | 34% (528)\ngit merge | 31% (482)\ngitk | 28% (404)\ngit rebase | 25% (365)\ngit stash | 22% (320)\n...................................\ngit remote update | 6% (92)\n\nAnalysis: strangely \"git add + git commit\" is slightly more used than\n\"git commit -a\"; I would suspect that \"git commit -a\" would dominate a\nbit over other forms of committing. What is for me more suprising is\nthat \"git commit ...\" has such large presence in often used\ncommands; I would think that it should be mostly used as 'sometimes'\ncommand.\n\nPity there was no \"git pull . \" to compare if people swith to\nnewer \"git merge \" for merging local branches.\n\n\nTop 10 sometimes used commands\n\n===================================\nCommand + options | sometimes\n-----------------------------------\ngit merge | 28% (443)\ngit pull | 27% (429)\ngit stash | 26% (380)\ngit remote | 26% (379)\ngit revert | 24% (346)\ngit rebase | 23% (332)\ngitk | 23% (328)\ngit fetch [] | 21% (331)\ngit commit --amend | 20% (314)\ngit commit ... | 20% (311)\n\nAnalysis: \"git commit --amend\" and \"git commit \" are here. That\nis understandable, and for me expected.\n\n\nTop 10 rarely used commands\n\n===================================\nCommand + options | rarely [%]\n-----------------------------------\ngit revert | 31% (454)\ngit pull | 22% (348)\ngit blame | 21% (328)\ngit fetch [] | 20% (313)\ngit rebase | 20% (291)\ngit remote | 20% (289)\ngit merge | 18% (286)\ngit apply | 18% (284)\ngit pull --rebase | 18% (283)\n\nAnalysis: \"git pull \", direct pull without using\nremote-tracking branches, is quite popular among rarely used\ncommands. Strange that \"git pull ---rebase\" is among rarely\nused; I would think it is question of workflow... but perhaps\nrarely encountered workflow.\n\nNotice that \"git blame\" is more 'rare' than gitk, which is in top 10\nof both 'often' and 'sometimes' table. I guess that this is caused by\nthe fact that git has other, often better, ways to examine and browse\nhistory.\n\n\nTop 12 never used (or not well known) commands\n\n==========================================================\nCommand + options never rarely sometime often\n----------------------------------------------------------\ngit relink 1217-84% 37- 3% 6- 0% 1- 0%\ngit cvsserver 1255-80% 23- 1% 8- 1% 3- 0%\ngit imap-send 1245-80% 35- 2% 6- 0% 2- 0%\ngit cvsexportcommit 1227-79% 26- 2% 16- 1% 12- 1%\ngit citool 1191-76% 51- 3% 15- 1% 17- 1%\ngit bundle 1184-76% 80- 5% 22- 1% 8- 1%\ngit request-pull 1110-76% 105- 7% 32- 2% 16- 1%\ngit instaweb 1142-73% 97- 6% 31- 2% 8- 1%\ngit ... --dirstat 1141-73% 93- 6% 39- 2% 8- 1%\ngit am -i 1129-72% 130- 8% 48- 3% 8- 1%\ngit blame -L 1123-72% 103- 7% 50- 3% 11- 1%\ngit daemon (pushing) 1110-71% 99- 6% 38- 2% 36- 2%\n\nIt is sorted by percentage; note that commands from different\nquestions might have higher percentage even though lower number of\nreplies: it is caused by the fact that number of replies (non skipped\nresponses) is different.\n\nAnalysis: Nobody is I think suprised by the commands that appear in\nthis table...\n\n\n\n29. Which of the following features do or did you use? \n30. Which features do you find unique and/or useful ones,\n compared to other systems (other SCMs)?\n (Choice - Multiple answers)\n\nThere were 1387 responses / 715 skipped for question 29., and\nthere were 1190 responses / 912 skipped for question 30.\n\nThese two questions were a bit related, but first deals with the\nfeatures one uses (even if they are not unique to Git, or very\ninteresting), while second deals with which features one think to be\nuseful (even if one doesn't use them).\n\nThe 'use' columns conatin percentage of replies for question 29.,\nwhile 'think' column is about question 30.\n\n==================================================\nFeature description use think\n--------------------------------------------------\ngitk or other history viewer 72% ---\nstash 56% 63%\nthe index --- 44%\nshell completion of commands 42% ---\nadd --interactive\n or other partial commit helper 35% 38%\ngit-gui or other commit tool 31% ---\ninteractive rebase 31% 39%\nsubmodules (subprojects) 30% 25%\nbisect 28% 37%\nmy own scripts / scriptability 25% 28%\ngit-aware shell prompt 23% ---\nintegration with IDE/editor 20% ---\nworking with dirty tree 18% 14%\nmergetool 17% 13%\nreflog 16% 16%\ndetaching HEAD 14% 9%\nnon-default hooks / hooks system 11% 19%\nshallow clone 9% 6%\neol conversion (crlf) 8% 6%\nseparate worktree 7% 11%\ngit-filter-branch 7% 7%\nalternates mechanism 6% 7%\ngitattributes 6% 4%\npatch management interface 5% ---\ngit bundle (off-line transport) 5% 10%\nmultiple worktrees (git-new-worktree) 3% ---\ncommit templates 2% 3%\n\nAbout items which are missing in one of columns: either it is a broad\nconcept so it is hard to say that one can \"use\" such feature, or\nfeature is not unique enough, or is not part of git... or I forgot to\nadd it to second part.\n\nAnalysis: most popular is using history viewer (such as gitk, qgit,\ngiggle, GitNub, git-cola, etc.); stash even though it is quite new\nfeature also is very popular. For me a bit strange is that partial\ncommits has more users than graphical commit tools like\ngit-gui/git-citool. Most (well, 44%) think that the index is unique\nand/or useful feature. Biggest difference in number of replies\nbetween 29. and 30. question belongs, from what I see, to bisect.\n\n\n31. If you use some important Git features not mentioned above, what\n are those? (Open ended text - Essay)\n\nTotal respondents: 115, skipped this question: 1987.\n\nThis I guess mean that we were exhaustive in our list of features,\nas it doesn't seem to be many omissions...\n\n-- \nJakub Narebski\nPoland\n"},{"id":"89958","messageId":"48C39145.9020709@op5.se","threadId":"15356","inReplyTo":"200809070017.09463.jnareb@gmail.com","subject":"Re: Git User's Survey 2008 partial summary, part 4 - how do we use Git","fromName":"Andreas Ericsson","fromEmail":"ae@op5.se","sentAt":"2008-09-07T08:31:01Z","receivedAt":"2008-09-07T08:31:01Z","isPatch":false,"sender":{"key":"ae@op5.se","avatar":"https://gravatar.com/avatar/426e89595c75a8f5252dd0c989e5fabe5bcac616e68557427ad9aef6b0ca342a?d=mp&s=160"},"body":"Jakub Narebski wrote:\n> \n> Analysis: strangely \"git add + git commit\" is slightly more used than\n> \"git commit -a\"; I would suspect that \"git commit -a\" would dominate a\n> bit over other forms of committing.\n\nI believe people have finally gotten to grasp with the index. Besides,\nthose answers are probably a bit skewed, as people that use git as just\na faster and better cvs/svn aren't likely to be those who answer the\nsurvey.\n\n> What is for me more suprising is\n> that \"git commit ...\" has such large presence in often used\n> commands; I would think that it should be mostly used as 'sometimes'\n> command.\n> \n\nI know myself and most others where I work use it as a 'fixup' command.\nOne commit is done, the second is being staged (git add or git add -i),\nwhile testing reveals some silly bug in one of the files from the last\ncommit. Assuming that file isn't touched again, it's easy to fix up\nusing \"git commit --amend \", as that automatically sets up a new\nindex so that the staged changes remain staged.\n\nThe alternative would be to do a mixed reset, git add , git commit\n--amend. Not so sexy.\n\n> \n> Top 10 rarely used commands\n> \n> ===================================\n> Command + options | rarely [%]\n> -----------------------------------\n> git revert | 31% (454)\n> git pull | 22% (348)\n> git blame | 21% (328)\n> git fetch [] | 20% (313)\n> git rebase | 20% (291)\n> git remote | 20% (289)\n> git merge | 18% (286)\n> git apply | 18% (284)\n> git pull --rebase | 18% (283)\n> \n> Analysis: \"git pull \", direct pull without using\n> remote-tracking branches, is quite popular among rarely used\n> commands. Strange that \"git pull ---rebase\" is among rarely\n> used; I would think it is question of workflow... but perhaps\n> rarely encountered workflow.\n> \n\nPerhaps many who use it don't know that they are. The \"git sync\"\nalias is much in use where I work. It's basically just a\n\"git pull --rebase && git push\". As it was suggested on the mailing\nlist and also listed in some tutorial somewhere (don't know where),\nI can imagine quite a lot of people using it without realising it.\n\n-- \nAndreas Ericsson andreas.ericsson@op5.se\nOP5 AB www.op5.se\nTel: +46 8-230225 Fax: +46 8-230231\n"},{"id":"89960","messageId":"200809071044.30804.jnareb@gmail.com","threadId":"15356","inReplyTo":"48C39145.9020709@op5.se","subject":"Re: Git User's Survey 2008 partial summary, part 4 - how do we use Git","fromName":"Jakub Narebski","fromEmail":"jnareb@gmail.com","sentAt":"2008-09-07T08:44:29Z","receivedAt":"2008-09-07T08:44:29Z","isPatch":false,"sender":{"key":"jnareb@gmail.com","avatar":"https://avatars.githubusercontent.com/u/2706?v=4"},"body":"Andreas Ericsson wrote:\n> Jakub Narebski wrote:\n>> \n>> Analysis: strangely \"git add + git commit\" is slightly more used than\n>> \"git commit -a\"; I would suspect that \"git commit -a\" would dominate a\n>> bit over other forms of committing.\n> \n> I believe people have finally gotten to grasp with the index. Besides,\n> those answers are probably a bit skewed, as people that use git as just\n> a faster and better cvs/svn aren't likely to be those who answer the\n> survey.\n\n\nFrom \"Git User's Survey 2008 partial summary, part 3\" subthread:\nMessage-Id: <200809060422.34779.jnareb@gmail.com>\n\n>> 10. Rate your own proficiency with Git\n>>\n>> Here again the distribution is almost perfectly symmetric.  Most\n>> people say that they are proficient with Git enough for everyday use\n>> (44%), slightly more (23%) can offer advice than selected 'casual,\n>> needs advice', while slightly more (7%) thinks themself novices than\n>> people who think that know Git very well (4%).\n>>\n>> For this survey it seems to imply that results would not be biased\n>> (skewed) by people who know Git well to very well.\n\nSo it doesn't seem that answers are skewed much...\n\n-- \nJakub Narebski\nPoland\n"},{"id":"90020","messageId":"2c6b72b30809071607m69148ec0mc9d0925e24bb7b7e@mail.gmail.com","threadId":"15356","inReplyTo":"20080906051530.GB28035@spearce.org","subject":"Re: Git User's Survey 2008 partial summary, part 3","fromName":"Jonas Fonseca","fromEmail":"jonas.fonseca@gmail.com","sentAt":"2008-09-07T23:07:50Z","receivedAt":"2008-09-07T23:07:50Z","isPatch":false,"sender":{"key":"jonas.fonseca@gmail.com","avatar":"https://gravatar.com/avatar/9b7fa23cce50269e5d164312b6ac5ae818a180f837b38f28f3bdf689dd7f96cd?d=mp&s=160"},"body":"On Sat, Sep 6, 2008 at 07:15, Shawn O. Pearce wrote:\n> Actually, I'm rather excited by what's going on in the Git IDE\n> space these days. The NetBeans plugin is seeing more activity,\n> and since GSoC so has EGit. Both projects have been steadily\n> picking up contributors. Quite a good thing.\n\nOne thing that I am wondering about is how to best integrate the\nvarious git-specific traits. NetBeans already has several VCS plugins,\nmany of which are derived from the same code and have been built to\nabide to the same UI design principles, and so naturally they have a\nsimilar \"feel\" to them. It seems foolish to not try and respect this\nwhen integrating git into the IDE space. Especially, to make it easier\nfor people to try out and switch to git.\n\nOne example is the question of how to enable users to work with the\nindex without adding too many obscure UI elements. NetBeans VCS\ninfrastructure and diff engine automatically annotates, which lines in\nthe editor has been changed, modified, or deleted, and allows users to\nrevert on a chunk level. While it seems very practical to extend the\nannotation bar to allow changes to be staged (I am not sure if this is\neven possible) it could end up cluttering a concept that many users\nare already very familiar with.\n\nI am curious what kind of considerations you have done on this topic\nin your work on EGit.\n\n-- \nJonas Fonseca\n"},{"id":"90026","messageId":"20080907234718.GC15460@spearce.org","threadId":"15356","inReplyTo":"2c6b72b30809071607m69148ec0mc9d0925e24bb7b7e@mail.gmail.com","subject":"Re: Git User's Survey 2008 partial summary, part 3","fromName":"Shawn O. Pearce","fromEmail":"spearce@spearce.org","sentAt":"2008-09-07T23:47:18Z","receivedAt":"2008-09-07T23:47:18Z","isPatch":false,"sender":{"key":"spearce@spearce.org","avatar":"https://avatars.githubusercontent.com/u/34844?v=4"},"body":"Jonas Fonseca wrote:\n> One thing that I am wondering about is how to best integrate the\n> various git-specific traits. NetBeans already has several VCS plugins,\n> many of which are derived from the same code and have been built to\n> abide to the same UI design principles, and so naturally they have a\n> similar \"feel\" to them. It seems foolish to not try and respect this\n> when integrating git into the IDE space. Especially, to make it easier\n> for people to try out and switch to git.\n\nYea, that's a big concern for any IDE implementor. Many of the\nEclipse team providers (if not all of them?) have derived from the\nCVS team provider that comes as part of the base platform. So they\noften share the same look-and-feel. But I think Eclipse defines\nless of a common VCS UI than what you are implying NetBeans does.\n\nIn Eclipse I think it is more common to define VCS custom views\n(little windows/toolbars) and let the user turn on/off each view\nas they see fit.\n \n> One example is the question of how to enable users to work with the\n> index without adding too many obscure UI elements. NetBeans VCS\n> infrastructure and diff engine automatically annotates, which lines in\n> the editor has been changed, modified, or deleted, and allows users to\n> revert on a chunk level. While it seems very practical to extend the\n> annotation bar to allow changes to be staged (I am not sure if this is\n> even possible) it could end up cluttering a concept that many users\n> are already very familiar with.\n\nYea, extending the annotation bar to show the index may actually\nbe useful quite useful. If its only 15 pixels of horizontal width\nlost that's actually worthwhile. I think an older version of EGit\nhad something like this until I broke it. I can't remember.\n\n> I am curious what kind of considerations you have done on this topic\n> in your work on EGit.\n\nWe've mostly ignored it. I haven't put a lot of thought into it\nbecause I have been focused on getting a solid JGit implementation\nin place.\n\nI do know that I really miss having an overview of what is\nessentially the left side of git-gui; the staged and unstaged\n(but modified) file lists. The resource decorator is nice, but it\ndoesn't quite give the same flat-tree overview.\n\n-- \nShawn.\n"},{"id":"90469","messageId":"200809112214.18366.jnareb@gmail.com","threadId":"15356","inReplyTo":"200809031607.19722.jnareb@gmail.com","subject":"Git User's Survey 2008 partial summary, part 5 - other SCM","fromName":"Jakub Narebski","fromEmail":"jnareb@gmail.com","sentAt":"2008-09-11T20:14:17Z","receivedAt":"2008-09-11T20:14:17Z","isPatch":false,"sender":{"key":"jnareb@gmail.com","avatar":"https://avatars.githubusercontent.com/u/2706?v=4"},"body":"This is partial summary of Git User's Survey 2008 for the state from\nSep 09, 2008, with almost 2300 responses (yes, _thousands_ responses).\nIt is based on \"Analysis\" page for this survey:\n  http://www.survs.com/shareResults?survey=M3PIVU72&rndm=OKJQ45LAG8\n\nWe have around 2295 individual responses (excluding 21 responses in the \n'testing' channel), for the time I got those data, as compared to\n683 individual responses for 2007 survey, and (I think)\n115 answers (Base = 115) for 2006 survey. That is a lot,...\nI wonder how many replies we would have at the end of survey, Oct 10?\n\n\n---\nBelow I concentrated on questions related to other SCM, although not all\nthose questions were from \"Other SCM\" section.\n\n12) What other SCM did or do you use?\n (Matrix - One answer per row)\n\nThe table below is sorted alphabetically, with exception of 'custom'\nand 'other' which are put at the end.\n\n==============================================================\nSCM | Never | Used it | Still use\n--------------------------------------------------------------\nAccuRev | 60% (1227) | 0% (6) | 0% (1)\nArch (or clone) | 54% (1108) | 7% (136) | 0% (1) 11\nBazaar-NG | 47% (969) | 12% (257) | 4% (76) 7\nBitKeeper | 55% (1135) | 5% (111) | 0% (3) 12\nClearCase | 53% (1101) | 7% (151) | 2% (37) 10\nCVS | 12% (242) | 62% (1277) | 13% (260) 1 \nDarcs | 48% (990) | 13% (262) | 2% (48) 6\nMercurial | 40% (819) | 18% (379) | 8% (174) 5\nMonotone | 55% (1136) | 5% (95) | 1% (16) 14\nMS SourceSafe | 46% (949) | 19% (401) | 1% (20) 3\nMS Studio Team System | 57% (1173) | 3% (56) | 0% (9)\nPerforce | 51% (1045) | 9% (191) | 3% (64) 9\nPVCS | 56% (1157) | 5% (100) | 0% (3) 13'\nRCS | 43% (881) | 19% (389) | 3% (59) 4\nSCCS | 56% (1146) | 5% (100) | 0% (7) 13'\nSubversion | 3% (59) | 35% (726) | 59% (1217) 2\nSVK | 51% (1043) | 11% (226) | 1% (15) 8\n..............................................................\ncustom (non-published) | 54% (1115) | 4% (90) | 1% (22) \nother | 51% (1042) | 5% (98) | 1% (22) \n--------------------------------------------------------------\nTotal respondents | 2060\nskipped this question | 235\n\nSide note: each of version control systems have at least one\nrespondent which still use it. That was not assured.\n\n\nTop 5 \"still use\", which should mean version control system which\nare used beside, and together with Git.\n\n=====================================\nSCM | still use\n-------------------------------------\n1. Subversion | 59% (1217)\n2. CVS | 13% (260)\n3. Mercurial | 8% (174)\n4. Bazaar-NG | 4% (76)\n5. Perforce | 3% (64)\n.....................................\n7. Darcs | 2% (48)\n10. Monotone | 1% (16)\n\nNote that 10th place for Monotone excludes 'custom' and 'other';\notherwise it would be 12th.\n\nAnalysis: Subversion (SVN) is most popular, eclipsing all other SCMs.\nThis might be caused by the fact that git-svn is very good in\nintegrating Subversion with Git enhances and emphasizes popularity of\nSubversion. CVS, at time very popular, leaves the field to Subversion\n(advertised as replacement fo CVS, as \"CVS done right\") and other\nsystems. This agrees with the number of people which have above\nversion control systems in Ohloh software stack [need new analysis for\nfinal version; for now you can use data from GitSurvey2007 page at git\nwiki], and with Debian Popularity Contest (popcon) [here also new data\nis needed].\n\nA bit suprising for me is high place of Perforce. Another strange\nthing (and a bit alarming) is that MS Visual SourceSafe has higher\nplace than Monotone; but that might be caused by different design and\ndifferent target groups of Monotone and Git, which might have caused\nthat the communities have almost no overlap; people choose either Git\nor Monotone, one or the other. BitKeeper has also a very low number\nof active users among Git users... but that is not that strange,\nconsidering history.\n\nSee also \"Adoption of various VCSes\" blog post by Newren:\nhttp://blogs.gnome.org/newren/2007/11/17/adoption-of-various-vcses/\nwhich is less than year old. This essay includes proprietary SCMs\nlike Perforce.\n\n\nTop 5 \"used it\" (assumption: no longer in use) version control\nsystems. Replaced by Git or other SCM, tried and abandoned, etc.\n\n=====================================\nSCM | used it\n-------------------------------------\n1. CVS | 62% (1277)\n2. Subversion | 35% (726)\n3. MS SourceSafe | 19% (401)\n4. RCS | 19% (389)\n5. Mercurial | 18% (379)\n.....................................\n6. Darcs | 13% (262)\n7. Bazaar-NG | 12% (257)\n14. Monotone | 5% (95)\n\nNote that again 'other' and 'custom' are excluded.\n\nAnalysis: We can see here former glory of CVS, which was once standard\nfor version control system. Its predecessor, single-file locking\nversion control system, RCS is also visible.\n\nA bit suprising is 3rd place of MS Visual SourceSafe; I really hope\nthat it is either \"I took a look at it and ran away\", or \"I had to use\nit but fortunately no longer\". And what do you think about this?\n\n\n13) Why did you choose/use Git? (if you use Git) \n What do you like about using Git? \n (Open ended text - Essay)\n\nTotal respondents 1723\nskipped this question 572\n\n14) Why did you choose/use other SCMs? (if you use other SCMs) \n What do you like about using other SCMs? \n (Open ended text - Essay)\n\nTotal respondents 1329\nskipped this question 966\n\n\n25) How do you publish/propagate your changes? \n (Choice - Multiple answers)\n\n======================================\nmethod | using it \n--------------------------------------\npush | 91% (1740)\npull request | 29% (551)\nformat-patch + email | 22% (414)\ngit bundle | 2% (39)\n......................................\ngit-svn | 26% (498)\nother foreign SCM^[*] | 2% (45)\n......................................\nother | 2% (32)\n======================================\nTotal respondents | 1904\nskipped this question | 391\n\nFootnote:\n=========\n[*] publishing to other foreing SCM:\n + Bazaar, using git-bzr\n + CVS, using git-cvsexportcommit\n using git-cvs (probably git-cvsserver)\n + ClearCase, using custom scripts\n unspecified method\n + Perforce, using git-p4\n using git-p4-import\n using patch\n + Mercurial, using fast-import (?)\n using email + hg mq\n + AccuRev, using git-acu\n + Quilt, using email\n + SourceSafe, using shared workdir\n using custom scripts\n + Subversion, using shared workdir (?)\n + SVK, using shared workdir\n + RCS, using custom scripts\n + TeamWare, using git-tw (unpublished yet)\n\n\nAnalysis, as pertinent to \"other SCM\" question: as you can see git-svn\nis very popular, just below pull request (which probably include\nGitHub social interface), and slightly higher (!) than format-patch\nplus email, i.e. patch-based workflow.\n\n\n26) If the way you publish your changes is not mentioned above,\n how do you publish your changes? Please explain. \n (Open ended text - Essay)\n\nTotal respondents 95\nskipped this question 2200\n\nHere are listed only methods which are truly not mentioned above\n(for example publishing to other SCMs usually doesn't count).\nI have also tried to remove duplicated entries:\n * git-bzr allows for bidirectional syncing with a git\n repository. Useful when a project is hosted in Bazaar.\n * Standard diff file encompassing all changes for a specific feature.\n * On one machine, I use rsync to push my repository upstream by hand\n (for HTTP access).\n * Personal use only (many responses).\n * we also use a script that uses various git=diff concoctions to\n generate patch bundles\n * format-patch + lighthouse;\n We create patches and attach to tickets in Lighthouse\n (lighthouseapp.com) for several open-source projects.\n * Patches on reviewboard\n * format-patch and attach to bugtracking system\n * Just so I can be a little more specific about how I publish. I\n primarily use stgit and then put the patches in an issue tracker\n (like trac).\n * I do format-patch. Then I've written a script that posts to\n pastebin.com My friend uses another script that downloads from\n pastebin.com ;)\n NOTE: why don't use gist.github.com, or even just GitHub...\n * forum or IRC to the project owners\n * format-patch and uploading to a webserver.\n * format-patch + save patch on wiki page\n * \"git diff > my.patch\" and upload to the drupal.org issue queue.\n * after-push hooks on github to campfire/lighthouse/twitter\n * GTP (Git Torrent Protocol -- my own implementation)\n * git archive\n * gitweb's snapshot/tarball plugin\n * Since my hosting provider is temperamental, I have many projects in\n a strange hierarchy, and I want to recreate my repository\n precisely, I just tar up my master folder and upload that.\n * Because of a retarded network setup out of my control, I use a\n disgusting script based on git clone --bare, tar, scp, ssh, and\n rsyncing.\n * Sometimes as simple and stupid as tarball / scp\n * Transport of entire repositories via CD/DVD.\n NOTE: wouldn't it be easier to use USB stick and bundles?\n * copy the whole changed file or tell them what to change since some\n windows people do not know what a patch is or how to apply it :-/\n * I use git for a repo with lots of huge files (RAW image files). git\n pull/push insists on packing up objects but I'd rather have one\n object per file in .git/objects, so I rsync those and then pull the\n changeset history via std means.\n * For all my projects, I set up a post-update hook on my dreamhost\n account to force update. I use a standard remote named \"stage\".\n * Package a RubyGem and upload it to the files area of the relevant\n RubyForge project, and make sure the same RubyGem is in my Git repo\n so GitHub will generate a gem as well.\n * Custom scripts for pull requests and patchbombing\n\nAn interesting comment:\n * At work I use git for tracking one project I work on. It is a\n project with Unix line endings in some files, and Windows file\n endings in the other. My company uses VSS which cannot handle Unix\n line endings and otherwise is not very powerful. So I use git to\n track my history, but use VSS as backup repository. Also I git pull\n my git repository to a backup repository on a network drive.\n\nAn explanation of choices:\n * There are cases where I will push to continuous integration\n repositories to get a specific tree tested. Otherwise it is all\n about pulling.\n * I do not publish my changes. Other people can access a copy of my\n code, but I do not use a git repository on a server. I do not\n understand that concept good enough (hint: that needs docs for\n total dummies)\n * I only use git for cloning/pulling repo's for now, since it is too\n hard to use. I don't have a lot of spare time to learn it since svn\n is easy and useful. I would use it if there was a doc that explains\n it to us svn people.\n\nAnd there is also:\n * For 27 and 28, there is no \"I don't know< I'm using the GUI's\",\n so I'm not ticking the answers\n * I don't know!\n * GitHub; Gitosis; Capistrano\n * Send them to somebody who has git access.\n NOTE: probably format-patch + email\n * can some one provide free private git service?\n ANSWER: There are many free _public_ git service, and a few\n non-free git services offer private repositories (e.g. GitHub)\n\n * Carrier pigeon\n\n\nAnalysis: as you can see format-patch + , where \nis not email or mailing list, but issue tracker, wiki, patch queue or\nother web site/web application is quite popular. \n\nSome people use different combinations of tar/zip and scp or other\ntransport, or rsync, or ssh to simply transfer the whole repository\n(some of which might be migitated by using bundles instead, but only\nif there is access to shell account on remote site and there is git on\nremote site to \"unpack\" bundle). Sometimes it is caused by the lack\nof decent features in a web server, sometimes it is lack of knowledge.\n\n\n15) Do you miss features in git that you know from other SCMs? \n If yes, what features are these (and from which SCM)? \n (Open ended text - Essay)\n\nTotal respondents 1046 (some/many of them wrote 'no')\nskipped this question 1249\n\nThis is just a very quick summary, based on a first few pages of\nresponses, Full analysis is I think best left for after closing the\nsurvey, because I think this would be a lot of work...\n\nSo here is preliminary list, or rather beginning of one:\n * sparse/partial checkout and clone (e.g. Perforce)\n * better MS Windows support (CVS, Mercurial, Darcs, Bazaar)\n * integration with IDE/RAD/editor/filemanager\n * good GUI/visual tools/graphical shell (on multiple platforms)\n * 'soft' locking aka. watching files (like in ClearCase)\n * equivalent of 'hg serve': fast web interface and pull-server\n in one (Mercurial)\n * better documentation (svnbook, hgbook, generic)\n * traversing history of individual file (unknown SCM)\n * better interface for staging individual changes\n (Darcs interface is better than 'git add -i')\n * better UI for submodules (less commands to issue, like in Subversion)\n * integrated conflict resolution tool\n * tracking empty repositories (Bazaar)\n + It would be helpful to have something like \"comments\" in monotone.\n + hackability and portability of Mercurial\n + good Unicode support, the current implicit handling sucks\n + Hg had less commands to do the same :) (Mercurial)\n + big file support (ClearCase) \n + maybe a 'info' style command where it lists all branches, tags,\n remotes and repo specific config. Rather than using separate\n commands.\n + I don't miss any *feature* from SVN, but I do miss some of the\n terse commands. Reverting my local changes to a file in Git always\n feels like hard work, and in SVN it's a dead simple command, etc.\n + svn:externals from subversion (especially the part where one can do\n partial imports from other repositories) something in the likes of\n hgforest, i.e. having 'live' submodules which are not fixed to a\n specific revision (from Mercurial)\n\n * better support for tracking third party tools like 'piston'\n for Subversion.\n NOTE: there is 'braid', formerly 'giston' for git\n * mainly fine-grained commit support (staging diff hunks),\n branch management, stashing\n NOTE: all of this is available in modern Git\n * easier to use commit identifiers. \"It's hard to memorize the\n seemingly random commit hash (even the short 7-character ones).\"\n\n\nNote that people used this question to put feature requests, instead\n_only_ things from other SCMs that they are missing in Git; and even\nwhen they did correctly, they forgot often to state _which_ SCM is\ngiven feature from.\n\n-- \nJakub Narebski\nPoland\n"},{"id":"90477","messageId":"3665a1a00809111403v1dfbef2cyf5377925ef087f2f@mail.gmail.com","threadId":"15356","inReplyTo":"200809112214.18366.jnareb@gmail.com","subject":"Re: Git User's Survey 2008 partial summary, part 5 - other SCM","fromName":"Anatol Pomozov","fromEmail":"anatol.pomozov@gmail.com","sentAt":"2008-09-11T21:03:11Z","receivedAt":"2008-09-11T21:03:11Z","isPatch":false,"sender":{"key":"anatol.pomozov@gmail.com","avatar":"https://gravatar.com/avatar/71fc20093402ce987148294ec0999d025209e52da6762789ff248cf5c317645f?d=mp&s=160"},"body":"Hi,\n> A bit suprising for me is high place of Perforce.\nSome huge companies use Perforce. A good example here is Google. But\nat the same time there are a lot of Google employees who interested in\ngit. Especially after Linus TechTalk. Googlers created a wrapper for\nperforce that tries to bring git to everyone and ~300 people actively\nuse it.\n--\nanatol\n"},{"id":"90484","messageId":"20080911215429.GC3187@coredump.intra.peff.net","threadId":"15356","inReplyTo":"200809112214.18366.jnareb@gmail.com","subject":"Re: Git User's Survey 2008 partial summary, part 5 - other SCM","fromName":"Jeff King","fromEmail":"peff@peff.net","sentAt":"2008-09-11T21:54:29Z","receivedAt":"2008-09-11T21:54:29Z","isPatch":false,"sender":{"key":"peff@peff.net","avatar":"https://avatars.githubusercontent.com/u/45925?v=4"},"body":"On Thu, Sep 11, 2008 at 10:14:17PM +0200, Jakub Narebski wrote:\n\n> SCM | Never | Used it | Still use\n> --------------------------------------------------------------\n[...]\n> CVS | 12% (242) | 62% (1277) | 13% (260) 1 \n\nI like to think of this as 1017 monitors that we have saved from getting\npunched in frustration.\n\n> 25) How do you publish/propagate your changes? \n> (Choice - Multiple answers)\n> \n> ======================================\n> method | using it \n> --------------------------------------\n> push | 91% (1740)\n\nObviously a quite popular choice. I wonder if this could be broken down\nmore in next year's survey:\n\n - push into my public repo, interested people pull\n - push into central repo\n\n-Peff\n"},{"id":"90486","messageId":"200809120005.34261.jnareb@gmail.com","threadId":"15356","inReplyTo":"48C98F92.40903@workspacewhiz.com","subject":"Re: Git User's Survey 2008 partial summary, part 5 - other SCM","fromName":"Jakub Narebski","fromEmail":"jnareb@gmail.com","sentAt":"2008-09-11T22:05:32Z","receivedAt":"2008-09-11T22:05:32Z","isPatch":false,"sender":{"key":"jnareb@gmail.com","avatar":"https://avatars.githubusercontent.com/u/2706?v=4"},"body":"Joshua Jensen wrote:\n> Anatol Pomozov wrote:\n>> Jakub Narebski wrote:\n>>>\n>>> A bit suprising for me is high place of Perforce.\n>>> \n>> Some huge companies use Perforce. A good example here is Google. But\n>> at the same time there are a lot of Google employees who interested in\n>> git. Especially after Linus TechTalk. Googlers created a wrapper for\n>> perforce that tries to bring git to everyone and ~300 people actively\n>> use it.\n>>\n> Many game companies use Perforce, as it has pretty good handling of \n> gigabytes upon gigabytes of art content. This is one area where a \n> centralized repository trumps the distributed ones with full history. \n> The checkout tree for one product I worked on was 70 gigabytes, and I \n> still mapped out a large number of files. You can imagine how that \n> becomes too unwieldy to have all history locally.\n\nDana How (Cc-ed) worked on better support in Git for very large and\nnot well compressed files (and usually with short history). Her\nsolution was to add support to git to limit size of blobs which\nentered packs (files/blobs of size larger than given limit were\nleft as loose objects), and share those loose objects using network\nfilesystem. You can search git mailing list to find exact details\nof this. IIRC mentioned patches were not accepted; instead there was\nproposed a solution where problematic blobs of very large files were\npacked into separate kept pack, and this kept pack (or packs, each\ncontaining single blob) were shared using some network filesystem\n(network drive).\n\n\nIn addition to that in Documentation/RelNotes-1.5.3 we have:\n\n - \"pack-objects\" honors \"delta\" attribute set in\n .gitattributes. It does not attempt to deltify blobs that\n come from paths with delta attribute set to false.\n\n(but as far as I can see this attribute is otherwise undocumented).\nThis can also help performance for a repository with very large files.\n\n-- \nJakub Narebski\nPoland\n"},{"id":"90491","messageId":"alpine.DEB.1.10.0809111543580.15169@asgard.lang.hm","threadId":"15356","inReplyTo":"200809112214.18366.jnareb@gmail.com","subject":"Re: Git User's Survey 2008 partial summary, part 5 - other SCM","fromName":"","fromEmail":"david@lang.hm","sentAt":"2008-09-11T22:51:38Z","receivedAt":"2008-09-11T22:51:38Z","isPatch":false,"sender":{"key":"david@lang.hm","avatar":null},"body":"On Thu, 11 Sep 2008, Jakub Narebski wrote:\n\n> A bit suprising for me is high place of Perforce. Another strange\n> thing (and a bit alarming) is that MS Visual SourceSafe has higher\n> place than Monotone; but that might be caused by different design and\n> different target groups of Monotone and Git, which might have caused\n> that the communities have almost no overlap; people choose either Git\n> or Monotone, one or the other. BitKeeper has also a very low number\n> of active users among Git users... but that is not that strange,\n> considering history.\n\nI think you are making the wrong assumption here.\n\nsomeone may use CVS becouse they contribute to a project that is only \navailabe via CVS\n\nsomeone may use Perforce becouse that is the VCS that their company uses\n\nif you do commercial development with windows you are almsot going to be \nrequired to use MS VSS.\n\nyou are looking at it from the point of view of 'which VCS would you \nselect for a new project', but that would be a very different question.\n\nI find it interesting that the number of people who use git and the other \nDVCS systems in so small. Is this becouse the 'market share' of those \nother systems is small? or becouse people who learn git aren't willing to \nput up with other systems (or vice-versa)? or is there some other trend \nor tendancy that makes people who select one DVCS more likely to work on \nsimilar projects, so people interested in those types of projects will \ngenerally just see a single DVCS system\n\nDavid Lang\n"},{"id":"90529","messageId":"200809121244.59067.jnareb@gmail.com","threadId":"15356","inReplyTo":"alpine.DEB.1.10.0809111543580.15169@asgard.lang.hm","subject":"Re: Git User's Survey 2008 partial summary, part 5 - other SCM","fromName":"Jakub Narebski","fromEmail":"jnareb@gmail.com","sentAt":"2008-09-12T10:44:58Z","receivedAt":"2008-09-12T10:44:58Z","isPatch":false,"sender":{"key":"jnareb@gmail.com","avatar":"https://avatars.githubusercontent.com/u/2706?v=4"},"body":"On Fri, 12 Sep 2008 00:51, david@lang.hm wrote:\n> On Thu, 11 Sep 2008, Jakub Narebski wrote:\n> \n> > A bit suprising for me is high place of Perforce. Another strange\n> > thing (and a bit alarming) is that MS Visual SourceSafe has higher\n> > place than Monotone; but that might be caused by different design and\n> > different target groups of Monotone and Git, which might have caused\n> > that the communities have almost no overlap; people choose either Git\n> > or Monotone, one or the other. BitKeeper has also a very low number\n> > of active users among Git users... but that is not that strange,\n> > considering history.\n> \n> I think you are making the wrong assumption here.\n> \n> Someone may use CVS becouse they contribute to a project that is only \n> availabe via CVS\n> \n> Someone may use Perforce becouse that is the VCS that their company uses\n\nTrue, I have forgot that \"I use this SCM\" (or \"I used this SCM\") doesn't\nnecessarily mean that one _choose_ this SCM. One can use some SCM\nbecause it is SCM project uses, or because their company requires it;\nbut not necessary, as git-svn and git-p4 show one can use Git, and\nmake it interact with respectively Subversion and Perforce, and trying\nto make it look like one uses this other SCM.\n\n> If you do commercial development with Windows you are almsot going to be \n> required to use MS VSS.\n\nI hope not. Even Microsoft themselves supposedly doesn't recommend\nSource(Un)Safe, but MS Visual Studio Team System.\n\n> You are looking at it from the point of view of 'which VCS would you \n> select for a new project', but that would be a very different question.\n\nRight.\n\n> I find it interesting that the number of people who use git and the other \n> DVCS systems in so small. Is this becouse the 'market share' of those \n> other systems is small? Or becouse people who learn git aren't willing to \n> put up with other systems (or vice-versa)? Or is there some other trend \n> or tendancy that makes people who select one DVCS more likely to work on \n> similar projects, so people interested in those types of projects will \n> generally just see a single DVCS system\n\nI don't think 59% (in the example case of using currently Subversion)\nis small. Take into account for example that there are people who (as\nseen from responses to other questions in this survey) use SCM (Git)\nonly to track their private work, never publishing. Then there are\npeople who do not track (perhaps with exception of web interfaces)\nother projects development using version control systems, even if they\ndo follow their development.\n\nSide note: the number of replies in \"still use\" category agrees with\nother data, like mentioned Ohloh stacks, or Debian popcon (package\npopularity), or Newren research, or vcscomparison research.\n\n-- \nJakub Narebski\nPoland\n"},{"id":"90646","messageId":"alpine.DEB.1.10.0809131408010.17867@asgard.lang.hm","threadId":"15356","inReplyTo":"200809121244.59067.jnareb@gmail.com","subject":"Re: Git User's Survey 2008 partial summary, part 5 - other SCM","fromName":"","fromEmail":"david@lang.hm","sentAt":"2008-09-13T21:11:29Z","receivedAt":"2008-09-13T21:11:29Z","isPatch":false,"sender":{"key":"david@lang.hm","avatar":null},"body":"On Fri, 12 Sep 2008, Jakub Narebski wrote:\n\n> On Fri, 12 Sep 2008 00:51, david@lang.hm wrote:\n>> On Thu, 11 Sep 2008, Jakub Narebski wrote:\n>>\n> True, I have forgot that \"I use this SCM\" (or \"I used this SCM\") doesn't\n> necessarily mean that one _choose_ this SCM. One can use some SCM\n> because it is SCM project uses, or because their company requires it;\n> but not necessary, as git-svn and git-p4 show one can use Git, and\n> make it interact with respectively Subversion and Perforce, and trying\n> to make it look like one uses this other SCM.\n\nI would expect people to still count those as using the other SCM. git-svn \nand git-p4 can do a lot, but they don't do everything, once in a while I \nwould expect to need to use the native commands for the upstream SCM\n\n>> I find it interesting that the number of people who use git and the other\n>> DVCS systems in so small. Is this becouse the 'market share' of those\n>> other systems is small? Or becouse people who learn git aren't willing to\n>> put up with other systems (or vice-versa)? Or is there some other trend\n>> or tendancy that makes people who select one DVCS more likely to work on\n>> similar projects, so people interested in those types of projects will\n>> generally just see a single DVCS system\n>\n> I don't think 59% (in the example case of using currently Subversion)\n> is small. Take into account for example that there are people who (as\n> seen from responses to other questions in this survey) use SCM (Git)\n> only to track their private work, never publishing. Then there are\n> people who do not track (perhaps with exception of web interfaces)\n> other projects development using version control systems, even if they\n> do follow their development.\n\nSubversion is not a Distributed SCM. I am pointing out the much smaller \noverlap between distributed SCM systems.\n\nDavid Lang\n"},{"id":"90648","messageId":"200809140003.29570.jnareb@gmail.com","threadId":"15356","inReplyTo":"alpine.DEB.1.10.0809131408010.17867@asgard.lang.hm","subject":"Re: Git User's Survey 2008 partial summary, part 5 - other SCM","fromName":"Jakub Narebski","fromEmail":"jnareb@gmail.com","sentAt":"2008-09-13T22:03:28Z","receivedAt":"2008-09-13T22:03:28Z","isPatch":false,"sender":{"key":"jnareb@gmail.com","avatar":"https://avatars.githubusercontent.com/u/2706?v=4"},"body":"On Sat, 13 Sep 2008, David Lang wrote:\n> On Fri, 12 Sep 2008, Jakub Narebski wrote:\n>> On Fri, 12 Sep 2008 00:51, david@lang.hm wrote:\n>>> On Thu, 11 Sep 2008, Jakub Narebski wrote:\n>>>\n>> True, I have forgot that \"I use this SCM\" (or \"I used this SCM\") doesn't\n>> necessarily mean that one _choose_ this SCM. One can use some SCM\n>> because it is SCM project uses, or because their company requires it;\n>> but not necessary, as git-svn and git-p4 show one can use Git, and\n>> make it interact with respectively Subversion and Perforce, and trying\n>> to make it look like one uses this other SCM.\n> \n> I would expect people to still count those as using the other SCM. git-svn \n> and git-p4 can do a lot, but they don't do everything, once in a while I \n> would expect to need to use the native commands for the upstream SCM\n\nWell, you can check it by analysing correlations between answers. \nUnfortunately there is no way to share raw data beside me generating it\nand sending, or you registering at Survs.com and me adding you as admin,\nbut you can emulate it using existing (just created) filters on\n\"Analyze\" page for Git User's Survey 2008:\n http://www.survs.com/shareResults?survey=M3PIVU72&rndm=OKJQ45LAG8\n http://tinyurl.com/GitSurvey2008Analyze\n\nFor example if you chose filter 'publish to git-svn', which select all\nreplies that have answered question 25.: how do you publish and checked\ngit-svn option, you can see that among those responders 6 of 555 (1%)\nselected that they _never_ used Subversion.\n\n>>> I find it interesting that the number of people who use git and the other\n>>> DVCS systems in so small. Is this becouse the 'market share' of those\n>>> other systems is small? Or becouse people who learn git aren't willing to\n>>> put up with other systems (or vice-versa)? Or is there some other trend\n>>> or tendancy that makes people who select one DVCS more likely to work on\n>>> similar projects, so people interested in those types of projects will\n>>> generally just see a single DVCS system\n>>\n>> I don't think 59% (in the example case of using currently Subversion)\n>> is small. Take into account for example that there are people who (as\n>> seen from responses to other questions in this survey) use SCM (Git)\n>> only to track their private work, never publishing. Then there are\n>> people who do not track (perhaps with exception of web interfaces)\n>> other projects development using version control systems, even if they\n>> do follow their development.\n> \n> Subversion is not a Distributed SCM. I am pointing out the much smaller \n> overlap between distributed SCM systems.\n\nOh, now I understand and agree with what you noticed: in \"I use\" most\nyou get for distributed SCM is 9%, most in \"used it\" is 19% (hmmm...\nis it really small?).\n\nFirst, from various nonscientific researches (Ohloh stack count, Debian\npopcon, counting well-known projects using given SCM; Google search\ntrends do not count) it looks like Git has biggest market share from\ndistributed version control systems (Subversion has larger share, but\nit is centralized). So one would expect at most\n /*100% \nas a percentage for sum of \"I used it\" and \"I use it\" columns, or at\nleast for \"I use it\" answer.\n\nSecond, one usually choose single SCM for his/her projects. And one\nusually interacts (for example tracks) projects which use the same SCM;\nkernel folks use Git (and kernel-related projects migrate to Git too,\ntake for example ALSA project), people who work more on Windows (Mozilla)\nand have product competing with Linux (OpenSolaris) ;-) use Mercurial,\nRuby and RoR people moved to Git, Ubuntu / Canonical use Bazaar-NG.\n\nThird different SCM have slightly different UI, and different underlying\nmodels. If you use one, usually you find other SCM concept and commands\nunintuitive (and of course vice versa).\n\nBut that is just speculation...\n-- \nJakub Narebski\nPoland\n"},{"id":"90657","messageId":"fcaeb9bf0809140345n11d41430ma9b4096c66776b0c@mail.gmail.com","threadId":"15356","inReplyTo":"200809112214.18366.jnareb@gmail.com","subject":"Re: Git User's Survey 2008 partial summary, part 5 - other SCM","fromName":"Nguyen Thai Ngoc Duy","fromEmail":"pclouds@gmail.com","sentAt":"2008-09-14T10:45:06Z","receivedAt":"2008-09-14T10:45:06Z","isPatch":false,"sender":{"key":"pclouds@gmail.com","avatar":"https://avatars.githubusercontent.com/u/720?v=4"},"body":"On 9/12/08, Jakub Narebski wrote:\n> 15) Do you miss features in git that you know from other SCMs?\n> If yes, what features are these (and from which SCM)?\n> (Open ended text - Essay)\n>\n> Total respondents 1046 (some/many of them wrote 'no')\n> skipped this question 1249\n>\n> This is just a very quick summary, based on a first few pages of\n> responses, Full analysis is I think best left for after closing the\n> survey, because I think this would be a lot of work...\n>\n> So here is preliminary list, or rather beginning of one:\n> * sparse/partial checkout and clone (e.g. Perforce)\n\nHave not read the survey result, but do you recall what is the most\nused term for sparse/partial checkout? What SCMs do sparse/partial\ncheckout? I think it could be usable as it is now in my\nwill-be-sent-again series, but I don't really know how people want it\nto from that.\n-- \nDuy\n"},{"id":"90675","messageId":"200809141532.58164.jnareb@gmail.com","threadId":"15356","inReplyTo":"fcaeb9bf0809140345n11d41430ma9b4096c66776b0c@mail.gmail.com","subject":"Re: Git User's Survey 2008 partial summary, part 5 - other SCM","fromName":"Jakub Narebski","fromEmail":"jnareb@gmail.com","sentAt":"2008-09-14T13:32:56Z","receivedAt":"2008-09-14T13:32:56Z","isPatch":false,"sender":{"key":"jnareb@gmail.com","avatar":"https://avatars.githubusercontent.com/u/2706?v=4"},"body":"On Sun, 14 Sep 2008, Nguyen Thai Ngoc Duy wrote:\n> On 9/12/08, Jakub Narebski wrote:\n> >\n> > 15) Do you miss features in git that you know from other SCMs?\n> > If yes, what features are these (and from which SCM)?\n> > (Open ended text - Essay)\n> >\n> > Total respondents 1046 (some/many of them wrote 'no')\n> > skipped this question 1249\n> >\n> > This is just a very quick summary, based on a first few pages of\n> > responses, Full analysis is I think best left for after closing the\n> > survey, because I think this would be a lot of work...\n> >\n> > So here is preliminary list, or rather beginning of one:\n> > * sparse/partial checkout and clone (e.g. Perforce)\n> \n> Have not read the survey result, but do you recall what is the most\n> used term for sparse/partial checkout? What SCMs do sparse/partial\n> checkout? I think it could be usable as it is now in my\n> will-be-sent-again series, but I don't really know how people want it\n> to from that.\n\nWell, I do not remember, and I think I haven't got original data now,\nonly the above summary. Besides, I haven't even began analysing data\nfrom question 34. about ideas for features, and I guess we would see\n'partial checkou' ide there too.\n\n>From short discussion on #git channel, Abhijit Menon-Sen (crab) said\nthat they are 'restricted views' in Perforce, but this terminology is\ncompletly alien to Git. Bjorn Steinbrink (doener_) mentioned that\nSubversion 1.5 acquired something they call 'sparse directories':\n\n http://blogs.open.collab.net/svn/2007/06/sparse-director.html\n\n(Sidenote: perhaps it would be worth to follow SVN UI for sparse \ncheckout here?). In short, people on #git agreed that 'partial \ncheckout' is a good name; I think that 'sparse checkout' better renders \nidea that one can checkout _set_ of files and directories, and it is \nnot only limited to single subdirectory ('subtree checkout' / \n'subdirectory checkout') or a single file.\n\nComparison table on Better SCM Initiative\n\n http://better-scm.berlios.de/comparison/comparison.html\n\nnames it \"Ability to Work only on One Directory of the Repository\" and \nuses '\"work_on_dir\"' id.\n\nHTH\n-- \nJakub Narebski\nPoland\n"},{"id":"90702","messageId":"alpine.DEB.1.10.0809142038190.19489@asgard.lang.hm","threadId":"15356","inReplyTo":"fcaeb9bf0809140345n11d41430ma9b4096c66776b0c@mail.gmail.com","subject":"Re: Git User's Survey 2008 partial summary, part 5 - other SCM","fromName":"","fromEmail":"david@lang.hm","sentAt":"2008-09-15T03:39:59Z","receivedAt":"2008-09-15T03:39:59Z","isPatch":false,"sender":{"key":"david@lang.hm","avatar":null},"body":"On Sun, 14 Sep 2008, Nguyen Thai Ngoc Duy wrote:\n\n> On 9/12/08, Jakub Narebski wrote:\n>> 15) Do you miss features in git that you know from other SCMs?\n>> If yes, what features are these (and from which SCM)?\n>> (Open ended text - Essay)\n>>\n>> Total respondents 1046 (some/many of them wrote 'no')\n>> skipped this question 1249\n>>\n>> This is just a very quick summary, based on a first few pages of\n>> responses, Full analysis is I think best left for after closing the\n>> survey, because I think this would be a lot of work...\n>>\n>> So here is preliminary list, or rather beginning of one:\n>> * sparse/partial checkout and clone (e.g. Perforce)\n>\n> Have not read the survey result, but do you recall what is the most\n> used term for sparse/partial checkout? What SCMs do sparse/partial\n> checkout? I think it could be usable as it is now in my\n> will-be-sent-again series, but I don't really know how people want it\n> to from that.\n\nthe most common use-case is people who want to follow a project but (at \nleast think that they) don't need the history. so they want to save \ntime/bandwideth/disk space by not downloading things they don't need.\n\nDavid Lang\n"},{"id":"90714","messageId":"48CE07FB.6010207@op5.se","threadId":"15356","inReplyTo":"alpine.DEB.1.10.0809142038190.19489@asgard.lang.hm","subject":"Re: Git User's Survey 2008 partial summary, part 5 - other SCM","fromName":"Andreas Ericsson","fromEmail":"ae@op5.se","sentAt":"2008-09-15T07:00:11Z","receivedAt":"2008-09-15T07:00:11Z","isPatch":false,"sender":{"key":"ae@op5.se","avatar":"https://gravatar.com/avatar/426e89595c75a8f5252dd0c989e5fabe5bcac616e68557427ad9aef6b0ca342a?d=mp&s=160"},"body":"david@lang.hm wrote:\n> On Sun, 14 Sep 2008, Nguyen Thai Ngoc Duy wrote:\n> \n>> On 9/12/08, Jakub Narebski wrote:\n>>> 15) Do you miss features in git that you know from other SCMs?\n>>> If yes, what features are these (and from which SCM)?\n>>> (Open ended text - Essay)\n>>>\n>>> Total respondents 1046 (some/many of them wrote 'no')\n>>> skipped this question 1249\n>>>\n>>> This is just a very quick summary, based on a first few pages of\n>>> responses, Full analysis is I think best left for after closing the\n>>> survey, because I think this would be a lot of work...\n>>>\n>>> So here is preliminary list, or rather beginning of one:\n>>> * sparse/partial checkout and clone (e.g. Perforce)\n>>\n>> Have not read the survey result, but do you recall what is the most\n>> used term for sparse/partial checkout? What SCMs do sparse/partial\n>> checkout? I think it could be usable as it is now in my\n>> will-be-sent-again series, but I don't really know how people want it\n>> to from that.\n> \n> the most common use-case is people who want to follow a project but (at \n> least think that they) don't need the history. so they want to save \n> time/bandwideth/disk space by not downloading things they don't need.\n> \n\nI disagree, and this would be a thin or shallow checkout to me (although\nI realize the nomenclature is unfortunately similar here). What I've\nheard is normally documentation writers that don't want to be able to\neven accidentally modify any code, so they just want Documentation/\nsubdirectory. Shallow clones are already supported (albeit with a lot\nof limitations).\n\n-- \nAndreas Ericsson andreas.ericsson@op5.se\nOP5 AB www.op5.se\nTel: +46 8-230225 Fax: +46 8-230231\n"},{"id":"90735","messageId":"20080915115114.GA19081@sirena.org.uk","threadId":"15356","inReplyTo":"200809121244.59067.jnareb@gmail.com","subject":"Re: Git User's Survey 2008 partial summary, part 5 - other SCM","fromName":"Mark Brown","fromEmail":"broonie@sirena.org.uk","sentAt":"2008-09-15T11:51:16Z","receivedAt":"2008-09-15T11:51:16Z","isPatch":false,"sender":{"key":"broonie@sirena.org.uk","avatar":"https://gravatar.com/avatar/9e798c729a4a709279df497d9608ad68422755c9670e92434a5436f7e607cf86?d=mp&s=160"},"body":"On Fri, Sep 12, 2008 at 12:44:58PM +0200, Jakub Narebski wrote:\n> On Fri, 12 Sep 2008 00:51, david@lang.hm wrote:\n\n> > If you do commercial development with Windows you are almsot going to be \n> > required to use MS VSS.\n\n> I hope not. Even Microsoft themselves supposedly doesn't recommend\n> Source(Un)Safe, but MS Visual Studio Team System.\n\nYes, it's strongly deprecated by Microsoft these days and has some known\nlimitations that aren't going to be addressed.\n\n-- \n\"You grabbed my hand and we fell into it, like a daydream - or a fever.\"\n"},{"id":"90848","messageId":"200809161912.06571.jnareb@gmail.com","threadId":"15356","inReplyTo":"alpine.DEB.1.10.0809142038190.19489@asgard.lang.hm","subject":"Re: Git User's Survey 2008 partial summary, part 5 - other SCM","fromName":"Jakub Narebski","fromEmail":"jnareb@gmail.com","sentAt":"2008-09-16T17:12:04Z","receivedAt":"2008-09-16T17:12:04Z","isPatch":false,"sender":{"key":"jnareb@gmail.com","avatar":"https://avatars.githubusercontent.com/u/2706?v=4"},"body":"On Mon, 15 Sep 2008, David Lang napisał:\n> On Sun, 14 Sep 2008, Nguyen Thai Ngoc Duy wrote:\n>> On 9/12/08, Jakub Narebski wrote:\n>>> 15) Do you miss features in git that you know from other SCMs?\n>>> If yes, what features are these (and from which SCM)?\n>>> (Open ended text - Essay)\n>>>\n>>> Total respondents 1046 (some/many of them wrote 'no')\n>>> skipped this question 1249\n>>>\n>>> This is just a very quick summary, based on a first few pages of\n>>> responses, Full analysis is I think best left for after closing the\n>>> survey, because I think this would be a lot of work...\n>>>\n>>> So here is preliminary list, or rather beginning of one:\n>>> * sparse/partial checkout and clone (e.g. Perforce)\n>>\n>> Have not read the survey result, but do you recall what is the most\n>> used term for sparse/partial checkout? What SCMs do sparse/partial\n>> checkout? I think it could be usable as it is now in my\n>> will-be-sent-again series, but I don't really know how people want it\n>> to from that.\n> \n> the most common use-case is people who want to follow a project but\n> (at least think that they) don't need the history. so they want to\n> save time/bandwideth/disk space by not downloading things they don't\n> need. \n\nErrr... the idea is to implement partial (sparse, narrow) _checkout_,\nwhich is doable, not partial/lazy _clone_, which is very, very \ndifficult.\n\nPartial checkout *might* help reduce disk space because usually \nrepository database (which is compressed and deltaified) is smaller \nthan checkout; and if it isn't you can always use _shallow_ clone (only \npart of history) and/or remote alternates (with reference repository\nnot on your quota, or on different filesystem). And it might help\ncheckout/commit/diff/status times (performance) on slow filesystems, \nwith extremely large number of files in full checkout. But contrary to\ncentralized SCM, where server is usually net away from working area,\npartial checkout doesn't affect network usage _at all_.\n\n-- \nJakub Narebski\nPoland\n"}]}