{"thread":{"id":"7634","subject":"cvsserver and wincvs (cvsgui)","startedAt":"2007-04-12T22:54:22Z","lastAt":"2007-04-13T03:11:19Z","messageCount":4,"participants":["Daniel Barkalow","Frank Lichtenheld","Jakub Narebski"],"isPatch":false,"patchVersion":null,"patchTotal":null},"messages":[{"id":"39240","messageId":"Pine.LNX.4.64.0704121821150.27922@iabervon.org","threadId":"7634","inReplyTo":null,"subject":"cvsserver and wincvs (cvsgui)","fromName":"Daniel Barkalow","fromEmail":"barkalow@iabervon.org","sentAt":"2007-04-12T22:54:22Z","receivedAt":"2007-04-12T22:54:22Z","isPatch":false,"sender":{"key":"barkalow@iabervon.org","avatar":"https://avatars.githubusercontent.com/u/55364219?v=4"},"body":"I'm having problems using git-cvsserver to support a wincvs user. It seems \nto be truncating binary files in both directions (e.g., we have a PNG that \nstarts, in hex, 89 50 4e 47 0d 0a ..., and it gets truncated between the \n0d and 0a), and it seems to fail up-to-date checks when checking in files \nthat are, in fact, up-to-date (and, so far as I can see, haven't been \nmodified anywhere else). I haven't tried much with this repository, but at \nleast the binary file commit and update worked for me with real cvs.\n\nThe user is using WinCvs 1.3.13.2 Beta 13 (Build 2), which seems to be the \nversion he could find at the time that would talk to both their local CVS \nserver and sourceforge.\n\nThe user doesn't have any of these problems talking to a real cvs server.\n\nAnybody got any experience with wincvs and cvsserver, or any idea how \nthings could be different between git-cvsserver and real cvs? He's \npresently going to use his own version control and just send us results, \nbut it would be nice if this could be magically worked out before we \nstarted working together closely again in a couple of weeks.\n\n\t-Daniel\n*This .sig left intentionally blank*\n"},{"id":"39248","messageId":"20070413000532.GI7085@planck.djpig.de","threadId":"7634","inReplyTo":"Pine.LNX.4.64.0704121821150.27922@iabervon.org","subject":"Re: cvsserver and wincvs (cvsgui)","fromName":"Frank Lichtenheld","fromEmail":"frank@lichtenheld.de","sentAt":"2007-04-13T00:05:33Z","receivedAt":"2007-04-13T00:05:33Z","isPatch":false,"sender":{"key":"frank@lichtenheld.de","avatar":"https://gravatar.com/avatar/b9f1d4b120e138f157c9e480d0818197c474628923786adb98f30017cdb99c3c?d=mp&s=160"},"body":"On Thu, Apr 12, 2007 at 06:54:22PM -0400, Daniel Barkalow wrote:\n> Anybody got any experience with wincvs and cvsserver, or any idea how \n> things could be different between git-cvsserver and real cvs? He's \n> presently going to use his own version control and just send us results, \n> but it would be nice if this could be magically worked out before we \n> started working together closely again in a couple of weeks.\n\nLooks like a problem with newline conversion. You should probably\ntry the gitcvs.allbinary option (available since 1.5.0 I think),\nthat makes git-cvsserver send all files with -kb which should suppress\nthe conversion.\n\nGruesse,\n-- \nFrank Lichtenheld \nwww: http://www.djpig.de/\n"},{"id":"39249","messageId":"evmhk0$p0l$1@sea.gmane.org","threadId":"7634","inReplyTo":"Pine.LNX.4.64.0704121821150.27922@iabervon.org","subject":"Re: cvsserver and wincvs (cvsgui)","fromName":"Jakub Narebski","fromEmail":"jnareb@gmail.com","sentAt":"2007-04-13T00:11:07Z","receivedAt":"2007-04-13T00:11:07Z","isPatch":false,"sender":{"key":"jnareb@gmail.com","avatar":"https://avatars.githubusercontent.com/u/2706?v=4"},"body":"[Cc: Daniel Barkalow , git@vger.kernel.org]\n\nDaniel Barkalow wrote:\n\n> I'm having problems using git-cvsserver to support a wincvs user. It seems \n> to be truncating binary files in both directions (e.g., we have a PNG that \n> starts, in hex, 89 50 4e 47 0d 0a ..., and it gets truncated between the \n> 0d and 0a), and it seems to fail up-to-date checks when checking in files \n> that are, in fact, up-to-date (and, so far as I can see, haven't been \n> modified anywhere else).\n\nWhich version of git do you use?\n\n\nIn git-cvsserner(1) (but not in git-config(1)) we have:\n\n The server should set the '-k' mode to binary when relevant, however,\n this is not really implemented yet. For now, you can force the server\n to set '-kb' for all files by setting the `gitcvs.allbinary` config\n variable. In proper GIT tradition, the contents of the files are\n always respected. No keyword expansion or newline munging is supported.\n\nDo you have gitcvs.allbinary set?\n\n\nBy the way, gitweb used to have the same problem. It was fixed by using\n:raw mode for outputting binary files, instead of set by default :utf8\n(see commit f5aa79d909f8c4838bf5).\n-- \nJakub Narebski\nWarsaw, Poland\nShadeHawk on #git\n"},{"id":"39255","messageId":"Pine.LNX.4.64.0704122309020.27922@iabervon.org","threadId":"7634","inReplyTo":"20070413000532.GI7085@planck.djpig.de","subject":"Re: cvsserver and wincvs (cvsgui)","fromName":"Daniel Barkalow","fromEmail":"barkalow@iabervon.org","sentAt":"2007-04-13T03:11:19Z","receivedAt":"2007-04-13T03:11:19Z","isPatch":false,"sender":{"key":"barkalow@iabervon.org","avatar":"https://avatars.githubusercontent.com/u/55364219?v=4"},"body":"On Fri, 13 Apr 2007, Frank Lichtenheld wrote:\n\n> On Thu, Apr 12, 2007 at 06:54:22PM -0400, Daniel Barkalow wrote:\n> > Anybody got any experience with wincvs and cvsserver, or any idea how \n> > things could be different between git-cvsserver and real cvs? He's \n> > presently going to use his own version control and just send us results, \n> > but it would be nice if this could be magically worked out before we \n> > started working together closely again in a couple of weeks.\n> \n> Looks like a problem with newline conversion. You should probably\n> try the gitcvs.allbinary option (available since 1.5.0 I think),\n> that makes git-cvsserver send all files with -kb which should suppress\n> the conversion.\n\nThat seems plausible; I think I'd read the documentation too while too \nsick and thought the fact that cvsserver, like git in general, doesn't \nchange anything meant that it would also tell clients that nothing should \nbe changed. Thanks; I've got a suitable version, and I'll try it and see \nif the user still has problems.\n\n\t-Daniel\n*This .sig left intentionally blank*\n"}]}