{"thread":{"id":"54990","subject":"add a blank line when a commit has no parent in log output?","startedAt":"2021-01-14T18:31:10Z","lastAt":"2021-01-25T07:09:58Z","messageCount":29,"participants":["Jason Pyeron","Philippe Blain","Junio C Hamano","Kyle Marek"],"isPatch":false,"patchVersion":null,"patchTotal":null},"messages":[{"id":"414372","messageId":"191201d6eaa3$4b585fa0$e2091ee0$@pdinc.us","threadId":"54990","inReplyTo":null,"subject":"add a blank line when a commit has no parent in log output?","fromName":"Jason Pyeron","fromEmail":"jpyeron@pdinc.us","sentAt":"2021-01-14T18:30:11Z","receivedAt":"2021-01-14T18:31:10Z","isPatch":false,"sender":{"key":"jpyeron@pdinc.us","avatar":"https://gravatar.com/avatar/c2e53452caa53d940768a1ffc9cf76196d851b9b534b7a39cd39852a70a0508f?d=mp&s=160"},"body":"Take this git log --format=\"%C(auto) %h% ad%d% s%C(green)% aE\" --graph --date=short\n\n| | | * 5505e019c2 2014-07-09 initial xxxxxx@xxxx\n| | | * 3e658f4085 2019-09-10 (wiki/wip-citest, origin/wip-citest) Added defau\n| | | * ad148aafe6 2019-09-10 Added default CI/CD Jenkinsfile (from f7daf088)\n\nOne might assume 5505e019c2 and 3e658f4085 are related. But git cat-file -p 5505e019c2\ntree 546c6b71f01e7fd086c8adb832518240b71a9075\nauthor sam swindell 1404878701 -0400\ncommitter sam swindell 1404878701 -0400\n\ninitial\n\n\nIs there a way to have it look like:\n\n| | | * 5505e019c2 2014-07-09 initial xxxxxx@xxxx\n| | |\n| | | * 3e658f4085 2019-09-10 (wiki/wip-citest, origin/wip-citest) Added defau\n| | | * ad148aafe6 2019-09-10 Added default CI/CD Jenkinsfile (from f7daf088)\n\nOr \n\n| | | # 5505e019c2 2014-07-09 initial xxxxxx@xxxx\n| | | * 3e658f4085 2019-09-10 (wiki/wip-citest, origin/wip-citest) Added defau\n| | | * ad148aafe6 2019-09-10 Added default CI/CD Jenkinsfile (from f7daf088)\n\nRespectfully,\n\nJason Pyeron\n\n\n--\nJason Pyeron | Architect\nPD Inc |\n10 w 24th St |\nBaltimore, MD |\n \n.mil: jason.j.pyeron.ctr...\n.com: jpyeron@pdinc.us\ntel : 202-741-9397\n\n\n\n"},{"id":"414375","messageId":"abc900c1-16cc-4ad4-4be3-c405924215cd@gmail.com","threadId":"54990","inReplyTo":"191201d6eaa3$4b585fa0$e2091ee0$@pdinc.us","subject":"Re: add a blank line when a commit has no parent in log output?","fromName":"Philippe Blain","fromEmail":"levraiphilippeblain@gmail.com","sentAt":"2021-01-14T19:29:27Z","receivedAt":"2021-01-14T19:30:27Z","isPatch":false,"sender":{"key":"levraiphilippeblain@gmail.com","avatar":"https://avatars.githubusercontent.com/u/44212482?v=4"},"body":"Hi Jason,\n\nLe 2021-01-14 à 13:30, Jason Pyeron a écrit :\n> Take this git log --format=\"%C(auto) %h% ad%d% s%C(green)% aE\" --graph --date=short\n> \n> | | | * 5505e019c2 2014-07-09 initial xxxxxx@xxxx\n> | | | * 3e658f4085 2019-09-10 (wiki/wip-citest, origin/wip-citest) Added defau\n> | | | * ad148aafe6 2019-09-10 Added default CI/CD Jenkinsfile (from f7daf088)\n> \n> One might assume 5505e019c2 and 3e658f4085 are related. But git cat-file -p 5505e019c2\n> tree 546c6b71f01e7fd086c8adb832518240b71a9075\n> author sam swindell 1404878701 -0400\n> committer sam swindell 1404878701 -0400\n> \n> initial\n> \n> \n> Is there a way to have it look like:\n> \n> | | | * 5505e019c2 2014-07-09 initial xxxxxx@xxxx\n> | | |\n> | | | * 3e658f4085 2019-09-10 (wiki/wip-citest, origin/wip-citest) Added defau\n> | | | * ad148aafe6 2019-09-10 Added default CI/CD Jenkinsfile (from f7daf088)\n> \n> Or\n> \n> | | | # 5505e019c2 2014-07-09 initial xxxxxx@xxxx\n> | | | * 3e658f4085 2019-09-10 (wiki/wip-citest, origin/wip-citest) Added defau\n> | | | * ad148aafe6 2019-09-10 Added default CI/CD Jenkinsfile (from f7daf088)\n> \n\nIf you remove '--graph', then you can add '--show-linear-break' [1]. Unfortunately\nthese two options do not work together. I think your suggestion to have the '*'\nbe changed to '#' for root commit is a great idea.\n\nIn the mean time, I use this trick:\n\n git log --date=short --format='%C(auto) %h% [%<(2,trunc)%p] ad%d% s%C(green)% aE'\n\nThis adds the abbreviated parent hashes (%p) but truncated to 2 characters ([2], [3]). So\nthe brackets will be empty for root commits.\n\nCheers,\n\nPhilippe.\n\n\n[1] https://git-scm.com/docs/git-log#Documentation/git-log.txt---show-linear-breakltbarriergt\n[2] https://git-scm.com/docs/git-log#Documentation/git-log.txt-empem\n[3] https://git-scm.com/docs/git-log#Documentation/git-log.txt-emltltNgttruncltruncmtruncem\n"},{"id":"414389","messageId":"196101d6eab6$20714550$6153cff0$@pdinc.us","threadId":"54990","inReplyTo":"abc900c1-16cc-4ad4-4be3-c405924215cd@gmail.com","subject":"RE: add a blank line when a commit has no parent in log output?","fromName":"Jason Pyeron","fromEmail":"jpyeron@pdinc.us","sentAt":"2021-01-14T20:44:59Z","receivedAt":"2021-01-14T20:45:37Z","isPatch":false,"sender":{"key":"jpyeron@pdinc.us","avatar":"https://gravatar.com/avatar/c2e53452caa53d940768a1ffc9cf76196d851b9b534b7a39cd39852a70a0508f?d=mp&s=160"},"body":"Kyle:\n\nNeed you to whip up a patch (back port it to current Cygwin git too), see below. It will help with cleaning up Cresaptown branches. Or if you think Watson can do it, give it to him.\n\n> -----Original Message-----\n> From: Philippe Blain \n> Sent: Thursday, January 14, 2021 2:29 PM\n> To: git@vger.kernel.org; Jason Pyeron \n> Subject: Re: add a blank line when a commit has no parent in log output?\n> \n> Hi Jason,\n> \n> Le 2021-01-14 à 13:30, Jason Pyeron a écrit :\n> > Take this git log --format=\"%C(auto) %h% ad%d% s%C(green)% aE\" --graph --date=short\n> >\n> > | | | * 5505e019c2 2014-07-09 initial xxxxxx@xxxx\n> > | | | * 3e658f4085 2019-09-10 (wiki/wip-citest, origin/wip-citest) Added defau\n> > | | | * ad148aafe6 2019-09-10 Added default CI/CD Jenkinsfile (from f7daf088)\n> >\n> > One might assume 5505e019c2 and 3e658f4085 are related. But git cat-file -p 5505e019c2\n> > tree 546c6b71f01e7fd086c8adb832518240b71a9075\n> > author sam swindell 1404878701 -0400\n> > committer sam swindell 1404878701 -0400\n> >\n> > initial\n> >\n> >\n> > Is there a way to have it look like:\n> >\n> > | | | * 5505e019c2 2014-07-09 initial xxxxxx@xxxx\n> > | | |\n> > | | | * 3e658f4085 2019-09-10 (wiki/wip-citest, origin/wip-citest) Added defau\n> > | | | * ad148aafe6 2019-09-10 Added default CI/CD Jenkinsfile (from f7daf088)\n> >\n> > Or\n> >\n> > | | | # 5505e019c2 2014-07-09 initial xxxxxx@xxxx\n> > | | | * 3e658f4085 2019-09-10 (wiki/wip-citest, origin/wip-citest) Added defau\n> > | | | * ad148aafe6 2019-09-10 Added default CI/CD Jenkinsfile (from f7daf088)\n> >\n> \n> If you remove '--graph', then you can add '--show-linear-break' [1]. Unfortunately\n> these two options do not work together. I think your suggestion to have the '*'\n> be changed to '#' for root commit is a great idea.\n\nPatch description\n\nWhen --graph is used\n\n--show-linear-break converts the * to a #\n\n--show-linear-break=x converts the * to a x\n\n> \n> In the mean time, I use this trick:\n> \n> git log --date=short --format='%C(auto) %h% [%<(2,trunc)%p] ad%d% s%C(green)% aE'\n> \n> This adds the abbreviated parent hashes (%p) but truncated to 2 characters ([2], [3]). So\n> the brackets will be empty for root commits.\n> \n> Cheers,\n> \n> Philippe.\n> \n> \n> [1] https://git-scm.com/docs/git-log#Documentation/git-log.txt---show-linear-breakltbarriergt\n> [2] https://git-scm.com/docs/git-log#Documentation/git-log.txt-empem\n> [3] https://git-scm.com/docs/git-log#Documentation/git-log.txt-emltltNgttruncltruncmtruncem\n\n"},{"id":"414412","messageId":"xmqq8s8vvw9m.fsf@gitster.c.googlers.com","threadId":"54990","inReplyTo":"191201d6eaa3$4b585fa0$e2091ee0$@pdinc.us","subject":"Re: add a blank line when a commit has no parent in log output?","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2021-01-15T01:12:05Z","receivedAt":"2021-01-15T01:13:10Z","isPatch":false,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"\"Jason Pyeron\" writes:\n\n> Is there a way to have it look like:\n>\n> | | | * 5505e019c2 2014-07-09 initial xxxxxx@xxxx\n> | | |\n> | | | * 3e658f4085 2019-09-10 (wiki/wip-citest, origin/wip-citest) Added defau\n> | | | * ad148aafe6 2019-09-10 Added default CI/CD Jenkinsfile (from f7daf088)\n>\n> Or \n>\n> | | | # 5505e019c2 2014-07-09 initial xxxxxx@xxxx\n> | | | * 3e658f4085 2019-09-10 (wiki/wip-citest, origin/wip-citest) Added defau\n> | | | * ad148aafe6 2019-09-10 Added default CI/CD Jenkinsfile (from f7daf088)\n\nThis latter variant won't work. Imagine we are showing --left-right\nfor example. Which side does '#' belong to?\n\nThe former is not so great in that it wastes a line, and the break\nwon't be as noticeable when --graph is *not* used with --oneline.\n\nIt would be great to show it more like this:\n\n | | | * 5505e019c2 2014-07-09 initial xxxxxx@xxxx\n | | | * 3e658f4085 2019-09-10 (wiki/wip-citest, origin/wip-citest) Added defau\n | | | * ad148aafe6 2019-09-10 Added default CI/CD Jenkinsfile (from f7daf088)\n\nThe point being that by shifting the column for the commit to the\nright, it shows that 5505 is not a child of 3e65 (and 3e65 is the\ntip of its lineage), and its parents do not appear in the displayed\nhistory. In the real life, the independent 'root' may be connected\nto the main history somehow, so you may see a graph like this:\n\n | | * 12345678 2021-01-14 merge xxxxx@xxxx into the history \n | | |\\\n | | | \\\n | | * \\ 23456789 2021-01-12 merge citest into the main history\n | | |\\ * 5505e019c2 2014-07-09 initial xxxxxx@xxxx\n | | | * 3e658f4085 2019-09-10 (wiki/wip-citest, origin/wip-citest) Added defau\n | | | * ad148aafe6 2019-09-10 Added default CI/CD Jenkinsfile (from f7daf088)\n\n\nHmm?\n"},{"id":"414547","messageId":"20210117110337.429994-2-kmarek@pdinc.us","threadId":"54990","inReplyTo":"20210117110337.429994-1-kmarek@pdinc.us","subject":"[PATCH 1/2] revision: Denote root commits with '#'","fromName":"Kyle Marek","fromEmail":"kmarek@pdinc.us","sentAt":"2021-01-17T11:03:36Z","receivedAt":"2021-01-17T12:07:03Z","isPatch":true,"sender":{"key":"kmarek@pdinc.us","avatar":null},"body":"This aids in identifying where an unrelated branch history starts when\nusing `git log --graph --oneline --all`\n\nSigned-off-by: Kyle Marek \n---\n revision.c | 6 ++++--\n 1 file changed, 4 insertions(+), 2 deletions(-)\n\ndiff --git a/revision.c b/revision.c\nindex 9dff845bed..8556923de8 100644\n--- a/revision.c\n+++ b/revision.c\n@@ -4191,9 +4191,11 @@ const char *get_revision_mark(const struct rev_info *revs, const struct commit *\n \t\t\treturn \"<\";\n \t\telse\n \t\t\treturn \">\";\n-\t} else if (revs->graph)\n+\t} else if (revs->graph) {\n+\t\tif (!commit->parents)\n+\t\t\treturn \"#\";\n \t\treturn \"*\";\n-\telse if (revs->cherry_mark)\n+\t} else if (revs->cherry_mark)\n \t\treturn \"+\";\n \treturn \"\";\n }\n-- \n2.29.2\n\n"},{"id":"414548","messageId":"20210117110337.429994-1-kmarek@pdinc.us","threadId":"54990","inReplyTo":"196101d6eab6$20714550$6153cff0$@pdinc.us","subject":"[PATCH 0/2] Option to modify revision mark for root commits","fromName":"Kyle Marek","fromEmail":"kmarek@pdinc.us","sentAt":"2021-01-17T11:03:35Z","receivedAt":"2021-01-17T12:07:08Z","isPatch":true,"sender":{"key":"kmarek@pdinc.us","avatar":null},"body":"This patch series allows --show-linear-break to be used with --graph,\nallowing for the revision mark to be changed for root commits.\n\nFeel free to squash away PATCH 1, or maybe even discard PATCH 2.\n\nJason: tested against Cygwin x86_64/release/git/git-2.30.0-1-src.tar.xz\n\nNote: PATCH 2 revision.c:2410 makes a second copy of optarg. This may\nnot be necessary.\n\nBackground:\n\nThe use case is --graph --oneline with unrelated histories. For example,\nin a hypothetical repository with an orphaned \"prebuilt\" branch\ncontaining builds of the master branch, the history may look like:\n\nkmarek@kyle-ppc64le /tmp/somerepo\n$ git log --graph --all --oneline\n* 02190b6 (prebuilt) add aarch64\n* 7b873f6 add x86_64\n* 26cc783 add ppc64le\n* 5b7186e (HEAD -> master) add Makefile\n* ea69093 implement cmdline parsing\n* a65df8a add main.c\n* 7727eb3 Initial commit\n\nAt first sight, the above log implies that 26cc783's parent is 5b7186e,\nor that master is an ancestor to prebuilt, but 26cc783 is the start of a\nnew history:\n\nkmarek@kyle-ppc64le /tmp/somerepo\n$ git log --graph --oneline master\n* 5b7186e (HEAD -> master) add Makefile\n* ea69093 implement cmdline parsing\n* a65df8a add main.c\n* 7727eb3 Initial commit\n\nkmarek@kyle-ppc64le /tmp/somerepo\n$ git log --graph --oneline prebuilt\n* 02190b6 (prebuilt) add aarch64\n* 7b873f6 add x86_64\n* 26cc783 add ppc64le\n\nTo identify the start of a new history:\n\nkmarek@kyle-ppc64le /tmp/somerepo\n$ git log --graph --all --oneline --show-linear-break\n* 02190b6 (prebuilt) add aarch64\n* 7b873f6 add x86_64\n# 26cc783 add ppc64le\n* 5b7186e (HEAD -> master) add Makefile\n* ea69093 implement cmdline parsing\n* a65df8a add main.c\n# 7727eb3 Initial commit\n\nkmarek@kyle-ppc64le /tmp/somerepo\n$ git log --graph --all --oneline --show-linear-break=I\n* 02190b6 (prebuilt) add aarch64\n* 7b873f6 add x86_64\nI 26cc783 add ppc64le\n* 5b7186e (HEAD -> master) add Makefile\n* ea69093 implement cmdline parsing\n* a65df8a add main.c\nI 7727eb3 Initial commit\n\nKyle Marek (2):\n revision: Denote root commits with '#'\n revision: implement --show-linear-break for --graph\n\n Documentation/rev-list-options.txt | 7 +++++++\n log-tree.c | 2 +-\n revision.c | 10 ++++++----\n revision.h | 1 +\n 4 files changed, 15 insertions(+), 5 deletions(-)\n\n-- \n2.29.2\n\n"},{"id":"414549","messageId":"20210117110337.429994-3-kmarek@pdinc.us","threadId":"54990","inReplyTo":"20210117110337.429994-1-kmarek@pdinc.us","subject":"[PATCH 2/2] revision: implement --show-linear-break for --graph","fromName":"Kyle Marek","fromEmail":"kmarek@pdinc.us","sentAt":"2021-01-17T11:03:37Z","receivedAt":"2021-01-17T12:07:30Z","isPatch":true,"sender":{"key":"kmarek@pdinc.us","avatar":null},"body":"where sets rev_info.break_revision_mark, the revision mark\nused for root commits.\n\nSigned-off-by: Kyle Marek \n---\n Documentation/rev-list-options.txt | 7 +++++++\n log-tree.c | 2 +-\n revision.c | 8 ++++----\n revision.h | 1 +\n 4 files changed, 13 insertions(+), 5 deletions(-)\n\ndiff --git a/Documentation/rev-list-options.txt b/Documentation/rev-list-options.txt\nindex 002379056a..93adb77c19 100644\n--- a/Documentation/rev-list-options.txt\n+++ b/Documentation/rev-list-options.txt\n@@ -1104,6 +1104,13 @@ This implies the `--topo-order` option by default, but the\n \tdo not belong to a linear branch. This option puts a barrier\n \tin between them in that case. If `` is specified, it\n \tis the string that will be shown instead of the default one.\n++\n+When --graph is used with --oneline, there is usually no vertical\n+space between commits, so the graph edge is not drawn. This can make\n+it hard to see that a history may end at one commit, while an\n+unrelated history starts at the next commit. This option changes the\n+revision mark for root commits. If `` is specified, it is\n+used as the new revision mark instead of the default one.\n \n ifdef::git-rev-list[]\n --count::\ndiff --git a/log-tree.c b/log-tree.c\nindex fd0dde97ec..f62300e404 100644\n--- a/log-tree.c\n+++ b/log-tree.c\n@@ -962,7 +962,7 @@ int log_tree_commit(struct rev_info *opt, struct commit *commit)\n \tif (opt->line_level_traverse)\n \t\treturn line_log_print(opt, commit);\n \n-\tif (opt->track_linear && !opt->linear && !opt->reverse_output_stage)\n+\tif (!opt->graph && opt->track_linear && !opt->linear && !opt->reverse_output_stage)\n \t\tfprintf(opt->diffopt.file, \"\\n%s\\n\", opt->break_bar);\n \tshown = log_tree_diff(opt, commit, &log);\n \tif (!shown && opt->loginfo && opt->always_show_header) {\ndiff --git a/revision.c b/revision.c\nindex 8556923de8..51deab2326 100644\n--- a/revision.c\n+++ b/revision.c\n@@ -2402,10 +2402,12 @@ static int handle_revision_opt(struct rev_info *revs, int argc, const char **arg\n \t\trevs->show_signature = 0;\n \t} else if (!strcmp(arg, \"--show-linear-break\")) {\n \t\trevs->break_bar = \" ..........\";\n+\t\trevs->break_revision_mark = \"#\";\n \t\trevs->track_linear = 1;\n \t\trevs->track_first_time = 1;\n \t} else if (skip_prefix(arg, \"--show-linear-break=\", &optarg)) {\n \t\trevs->break_bar = xstrdup(optarg);\n+\t\trevs->break_revision_mark = xstrdup(optarg);\n \t\trevs->track_linear = 1;\n \t\trevs->track_first_time = 1;\n \t} else if (skip_prefix(arg, \"--show-notes=\", &optarg) ||\n@@ -2530,8 +2532,6 @@ static int handle_revision_opt(struct rev_info *revs, int argc, const char **arg\n \t\t\tunkv[(*unkc)++] = arg;\n \t\treturn opts;\n \t}\n-\tif (revs->graph && revs->track_linear)\n-\t\tdie(\"--show-linear-break and --graph are incompatible\");\n \n \treturn 1;\n }\n@@ -4192,8 +4192,8 @@ const char *get_revision_mark(const struct rev_info *revs, const struct commit *\n \t\telse\n \t\t\treturn \">\";\n \t} else if (revs->graph) {\n-\t\tif (!commit->parents)\n-\t\t\treturn \"#\";\n+\t\tif (revs->break_revision_mark && !commit->parents)\n+\t\t\treturn revs->break_revision_mark;\n \t\treturn \"*\";\n \t} else if (revs->cherry_mark)\n \t\treturn \"+\";\ndiff --git a/revision.h b/revision.h\nindex 086ff10280..83b2ecef56 100644\n--- a/revision.h\n+++ b/revision.h\n@@ -297,6 +297,7 @@ struct rev_info {\n \n \tstruct commit_list *previous_parents;\n \tconst char *break_bar;\n+\tconst char *break_revision_mark;\n \n \tstruct revision_sources *sources;\n \n-- \n2.29.2\n\n"},{"id":"414558","messageId":"xmqq7dobmfrq.fsf@gitster.c.googlers.com","threadId":"54990","inReplyTo":"20210117110337.429994-2-kmarek@pdinc.us","subject":"Re: [PATCH 1/2] revision: Denote root commits with '#'","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2021-01-17T21:10:01Z","receivedAt":"2021-01-17T21:11:12Z","isPatch":true,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Kyle Marek writes:\n\n> This aids in identifying where an unrelated branch history starts when\n> using `git log --graph --oneline --all`\n>\n> Signed-off-by: Kyle Marek \n> ---\n> revision.c | 6 ++++--\n> 1 file changed, 4 insertions(+), 2 deletions(-)\n\nNo tests?\n\n> diff --git a/revision.c b/revision.c\n> index 9dff845bed..8556923de8 100644\n> --- a/revision.c\n> +++ b/revision.c\n> @@ -4191,9 +4191,11 @@ const char *get_revision_mark(const struct rev_info *revs, const struct commit *\n> \t\t\treturn \"<\";\n> \t\telse\n> \t\t\treturn \">\";\n> -\t} else if (revs->graph)\n> +\t} else if (revs->graph) {\n> +\t\tif (!commit->parents)\n> +\t\t\treturn \"#\";\n> \t\treturn \"*\";\n> -\telse if (revs->cherry_mark)\n> +\t} else if (revs->cherry_mark)\n> \t\treturn \"+\";\n> \treturn \"\";\n> }\n\nHere is what I tried to come up with, but somehow the \"#\" marker is\nnot showing for me.\n\nThe \"counted plus --left-right\" tests stress why a single \"#\" is not\ngood enough. I think the patch also needs to replace \"<\" and \">\"\nfor root commits that are left and right---in the tests, I used \"L\"\nto denote \"root that is on the left side\" (and \"R\" for the right\nside) instead of single \"#\", so that we do not to lose information.\n\nBy the way, as I already said in the original thread, I do not think\nthe '#' marking is a good idea; I'd rather see the root commit shown\nby shifting columns.\n\nAnyway, here is to test [1/2].\n\n t/t6020-rev-list-boundary.sh | 132 +++++++++++++++++++++++++++++++++++++++++++\n 1 file changed, 132 insertions(+)\n\ndiff --git i/t/t6020-rev-list-boundary.sh w/t/t6020-rev-list-boundary.sh\nnew file mode 100755\nindex 0000000000..f25e041951\n--- /dev/null\n+++ w/t/t6020-rev-list-boundary.sh\n@@ -0,0 +1,132 @@\n+#!/bin/sh\n+\n+test_description='rev-list/log boundary and root'\n+\n+. ./test-lib.sh\n+\n+test_expect_success 'setup' '\n+\ttest_commit A &&\n+\ttest_commit B &&\n+\tgit reset --hard A &&\n+\ttest_commit C &&\n+\n+\tgit checkout --orphan side &&\n+\tgit rm -fr . &&\n+\ttest_commit X &&\n+\ttest_commit Y &&\n+\n+\ttest_tick && git merge --allow-unrelated-histories -m \"M\" B &&\n+\ttest_tick && git merge -m \"N\" C &&\n+\ttest_commit Z\n+'\n+\n+test_expect_success 'log with boundary' '\n+\tgit log --graph --boundary --format='%s' ^A ^X Z >actual &&\n+\tsed -e \"s/Q$//\" >expect <<-\\EOF &&\n+\t* Z\n+\t* N\n+\t|\\ Q\n+\t| * C\n+\t* | M\n+\t|\\ \\ Q\n+\t| * | B\n+\t| |/ Q\n+\t* | Y\n+\to | X\n+\t / Q\n+\to A\n+\tEOF\n+\ttest_cmp expect actual\n+'\n+\n+test_expect_success 'log --left-right with symmetric boundary' '\n+\tgit log --graph --left-right --boundary --format='%s' B...C >actual &&\n+\tsed -e \"s/Q$//\" >expect <<-\\EOF &&\n+\t> C\n+\t| < B\n+\t|/ Q\n+\to A\n+\tEOF\n+\ttest_cmp expect actual\n+'\n+\n+test_expect_success 'log --left-right with asymmetric boundary' '\n+\tgit log --graph --left-right --boundary --format='%s' ^A ^X Z >actual &&\n+\tsed -e \"s/Q$//\" >expect <<-\\EOF &&\n+\t> Z\n+\t> N\n+\t|\\ Q\n+\t| > C\n+\t> | M\n+\t|\\ \\ Q\n+\t| > | B\n+\t| |/ Q\n+\t> | Y\n+\to | X\n+\t / Q\n+\to A\n+\tEOF\n+\ttest_cmp expect actual\n+'\n+\n+test_expect_failure 'log down to root' '\n+\tgit log --graph --format='%s' Z >actual &&\n+\tsed -e \"s/Q$//\" >expect <<-\\EOF &&\n+\t* Z\n+\t* N\n+\t|\\ Q\n+\t| * C\n+\t* | M\n+\t|\\ \\ Q\n+\t| * | B\n+\t| |/ Q\n+\t| # A\n+\t* Y\n+\t# X\n+\tEOF\n+\ttest_cmp expect actual\n+'\n+\n+test_expect_failure 'log down to root' '\n+\tgit log --graph --format='%s' B Y >actual &&\n+\tsed -e \"s/Q$//\" >expect <<-\\EOF &&\n+\t* Y\n+\t# X\n+\t* B\n+\t# A\n+\tEOF\n+\ttest_cmp expect actual\n+'\n+\n+test_expect_failure 'log that happens to show root' '\n+\tgit log --graph -3 --format='%s' B Y >actual &&\n+\tsed -e \"s/Q$//\" >expect <<-\\EOF &&\n+\t* Y\n+\t# X\n+\t* B\n+\tEOF\n+\ttest_cmp expect actual\n+'\n+\n+test_expect_failure 'log --left-right down to root' '\n+\tgit log --graph --left-right --format='%s' B...Y >actual &&\n+\tsed -e \"s/Q$//\" >expect <<-\\EOF &&\n+\t> Y\n+\tR X\n+\t< B\n+\tL A\n+\tEOF\n+\ttest_cmp expect actual\n+'\n+\n+test_expect_failure 'log --left-right that happens to show root' '\n+\tgit log --graph -3 --left-right --format='%s' B...Y >actual &&\n+\tsed -e \"s/Q$//\" >expect <<-\\EOF &&\n+\t> Y\n+\tR X\n+\t< B\n+\tEOF\n+\ttest_cmp expect actual\n+'\n+\n+test_done\n"},{"id":"414561","messageId":"xmqq35yzmbf3.fsf@gitster.c.googlers.com","threadId":"54990","inReplyTo":"20210117110337.429994-2-kmarek@pdinc.us","subject":"Re: [PATCH 1/2] revision: Denote root commits with '#'","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2021-01-17T22:44:00Z","receivedAt":"2021-01-17T22:44:50Z","isPatch":true,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Kyle Marek writes:\n\n> Subject: Re: [PATCH 1/2] revision: Denote root commits with '#'\n\nDowncase \"D\"; this will stand out in \"git shortlog --no-merges\" for\na wrong reason otherwise.\n\n> This aids in identifying where an unrelated branch history starts when\n> using `git log --graph --oneline --all`\n\nThis is triggerd only with --show-linear-break option, when combined\nwith [2/2]? I think that is a bug introduced in the next step.\n\n> Signed-off-by: Kyle Marek \n> ---\n> revision.c | 6 ++++--\n> 1 file changed, 4 insertions(+), 2 deletions(-)\n>\n> diff --git a/revision.c b/revision.c\n> index 9dff845bed..8556923de8 100644\n> --- a/revision.c\n> +++ b/revision.c\n> @@ -4191,9 +4191,11 @@ const char *get_revision_mark(const struct rev_info *revs, const struct commit *\n> \t\t\treturn \"<\";\n> \t\telse\n> \t\t\treturn \">\";\n> -\t} else if (revs->graph)\n> +\t} else if (revs->graph) {\n> +\t\tif (!commit->parents)\n> +\t\t\treturn \"#\";\n> \t\treturn \"*\";\n> -\telse if (revs->cherry_mark)\n> +\t} else if (revs->cherry_mark)\n> \t\treturn \"+\";\n> \treturn \"\";\n> }\n"},{"id":"414562","messageId":"xmqqsg6zkwa8.fsf@gitster.c.googlers.com","threadId":"54990","inReplyTo":"20210117110337.429994-3-kmarek@pdinc.us","subject":"Re: [PATCH 2/2] revision: implement --show-linear-break for --graph","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2021-01-17T22:56:15Z","receivedAt":"2021-01-17T22:57:17Z","isPatch":true,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Kyle Marek writes:\n\n> where sets rev_info.break_revision_mark, the revision mark\n> used for root commits.\n\nPlease make sure that the body of the proposed log message begins\nwith a full sentence, not as a continuation of a sentence that the\ntitle started (as a consequence, the title must be understandable\nwithout the help of the beginning part of the body, too).\n\n> Signed-off-by: Kyle Marek \n> ---\n\n> diff --git a/revision.c b/revision.c\n> index 8556923de8..51deab2326 100644\n> --- a/revision.c\n> +++ b/revision.c\n> @@ -2402,10 +2402,12 @@ static int handle_revision_opt(struct rev_info *revs, int argc, const char **arg\n> \t\trevs->show_signature = 0;\n> \t} else if (!strcmp(arg, \"--show-linear-break\")) {\n> \t\trevs->break_bar = \" ..........\";\n> +\t\trevs->break_revision_mark = \"#\";\n> \t\trevs->track_linear = 1;\n> \t\trevs->track_first_time = 1;\n> \t} else if (skip_prefix(arg, \"--show-linear-break=\", &optarg)) {\n> \t\trevs->break_bar = xstrdup(optarg);\n> +\t\trevs->break_revision_mark = xstrdup(optarg);\n> \t\trevs->track_linear = 1;\n> \t\trevs->track_first_time = 1;\n> \t} else if (skip_prefix(arg, \"--show-notes=\", &optarg) ||\n\nIn other words, revs->break_revision_mark is left NULL unless\n--show-linear-break is given.\n\n> @@ -4192,8 +4192,8 @@ const char *get_revision_mark(const struct rev_info *revs, const struct commit *\n> \t\telse\n> \t\t\treturn \">\";\n> \t} else if (revs->graph) {\n> -\t\tif (!commit->parents)\n> -\t\t\treturn \"#\";\n> +\t\tif (revs->break_revision_mark && !commit->parents)\n> +\t\t\treturn revs->break_revision_mark;\n\nAnd that causes this to break. Now \"--graph\" alone won't show '#'\nfor the root commits, despite that is what [1/2] wanted to do.\n\nHere is a fix-up, plus some minimum tests. \n\nThe part to teach left-right codepath to show L/R is a fix-up to\n[1/2], not to this step. You might want to change them to some\nleft/right punctuation letters, like () or [].\n\nThe other hunks in revision.c are fixes to step [2/2].\n\nI didn't test a custom --show-linear-break='My break line' in the\nattachedtest, so that it can be squashed into your [1/2] to test the\nfeature that step adds. You should be able to add tests for that\nfeature in this step [2/2] on top.\n\nI still am skeptical that spending 3 more letters to denote roots is\nworth it, though.\n\n revision.c | 11 ++--\n t/t6020-rev-list-boundary.sh | 132 +++++++++++++++++++++++++++++++++++++++++++\n 2 files changed, 138 insertions(+), 5 deletions(-)\n\ndiff --git c/revision.c w/revision.c\nindex 33fbef5c08..55521c53af 100644\n--- c/revision.c\n+++ w/revision.c\n@@ -2402,7 +2402,6 @@ static int handle_revision_opt(struct rev_info *revs, int argc, const char **arg\n \t\trevs->show_signature = 0;\n \t} else if (!strcmp(arg, \"--show-linear-break\")) {\n \t\trevs->break_bar = \" ..........\";\n-\t\trevs->break_revision_mark = \"#\";\n \t\trevs->track_linear = 1;\n \t\trevs->track_first_time = 1;\n \t} else if (skip_prefix(arg, \"--show-linear-break=\", &optarg)) {\n@@ -4219,12 +4218,14 @@ const char *get_revision_mark(const struct rev_info *revs, const struct commit *\n \t\treturn \"=\";\n \telse if (!revs || revs->left_right) {\n \t\tif (commit->object.flags & SYMMETRIC_LEFT)\n-\t\t\treturn \"<\";\n+\t\t\treturn commit->parents ? \"<\" : \"L\";\n \t\telse\n-\t\t\treturn \">\";\n+\t\t\treturn commit->parents ? \">\" : \"R\";\n \t} else if (revs->graph) {\n-\t\tif (revs->break_revision_mark && !commit->parents)\n-\t\t\treturn revs->break_revision_mark;\n+\t\tif (!commit->parents)\n+\t\t\treturn (revs->break_revision_mark \n+\t\t\t\t? revs->break_revision_mark\n+\t\t\t\t: \"#\");\n \t\treturn \"*\";\n \t} else if (revs->cherry_mark)\n \t\treturn \"+\";\ndiff --git c/t/t6020-rev-list-boundary.sh w/t/t6020-rev-list-boundary.sh\nnew file mode 100755\nindex 0000000000..35614e9baf\n--- /dev/null\n+++ w/t/t6020-rev-list-boundary.sh\n@@ -0,0 +1,132 @@\n+#!/bin/sh\n+\n+test_description='rev-list/log boundary and root'\n+\n+. ./test-lib.sh\n+\n+test_expect_success 'setup' '\n+\ttest_commit A &&\n+\ttest_commit B &&\n+\tgit reset --hard A &&\n+\ttest_commit C &&\n+\n+\tgit checkout --orphan side &&\n+\tgit rm -fr . &&\n+\ttest_commit X &&\n+\ttest_commit Y &&\n+\n+\ttest_tick && git merge --allow-unrelated-histories -m \"M\" B &&\n+\ttest_tick && git merge -m \"N\" C &&\n+\ttest_commit Z\n+'\n+\n+test_expect_success 'log with boundary' '\n+\tgit log --graph --boundary --format='%s' ^A ^X Z >actual &&\n+\tsed -e \"s/Q$//\" >expect <<-\\EOF &&\n+\t* Z\n+\t* N\n+\t|\\ Q\n+\t| * C\n+\t* | M\n+\t|\\ \\ Q\n+\t| * | B\n+\t| |/ Q\n+\t* | Y\n+\to | X\n+\t / Q\n+\to A\n+\tEOF\n+\ttest_cmp expect actual\n+'\n+\n+test_expect_success 'log --left-right with symmetric boundary' '\n+\tgit log --graph --left-right --boundary --format='%s' B...C >actual &&\n+\tsed -e \"s/Q$//\" >expect <<-\\EOF &&\n+\t> C\n+\t| < B\n+\t|/ Q\n+\to A\n+\tEOF\n+\ttest_cmp expect actual\n+'\n+\n+test_expect_success 'log --left-right with asymmetric boundary' '\n+\tgit log --graph --left-right --boundary --format='%s' ^A ^X Z >actual &&\n+\tsed -e \"s/Q$//\" >expect <<-\\EOF &&\n+\t> Z\n+\t> N\n+\t|\\ Q\n+\t| > C\n+\t> | M\n+\t|\\ \\ Q\n+\t| > | B\n+\t| |/ Q\n+\t> | Y\n+\to | X\n+\t / Q\n+\to A\n+\tEOF\n+\ttest_cmp expect actual\n+'\n+\n+test_expect_success 'log down to root' '\n+\tgit log --graph --format='%s' Z >actual &&\n+\tsed -e \"s/Q$//\" >expect <<-\\EOF &&\n+\t* Z\n+\t* N\n+\t|\\ Q\n+\t| * C\n+\t* | M\n+\t|\\ \\ Q\n+\t| * | B\n+\t| |/ Q\n+\t| # A\n+\t* Y\n+\t# X\n+\tEOF\n+\ttest_cmp expect actual\n+'\n+\n+test_expect_success 'log down to root' '\n+\tgit log --graph --format='%s' B Y >actual &&\n+\tsed -e \"s/Q$//\" >expect <<-\\EOF &&\n+\t* Y\n+\t# X\n+\t* B\n+\t# A\n+\tEOF\n+\ttest_cmp expect actual\n+'\n+\n+test_expect_success 'log that happens to show root' '\n+\tgit log --graph -3 --format='%s' B Y >actual &&\n+\tsed -e \"s/Q$//\" >expect <<-\\EOF &&\n+\t* Y\n+\t# X\n+\t* B\n+\tEOF\n+\ttest_cmp expect actual\n+'\n+\n+test_expect_success 'log --left-right down to root' '\n+\tgit log --graph --left-right --format='%s' B...Y >actual &&\n+\tsed -e \"s/Q$//\" >expect <<-\\EOF &&\n+\t> Y\n+\tR X\n+\t< B\n+\tL A\n+\tEOF\n+\ttest_cmp expect actual\n+'\n+\n+test_expect_success 'log --left-right that happens to show root' '\n+\tgit log --graph -3 --left-right --format='%s' B...Y >actual &&\n+\tsed -e \"s/Q$//\" >expect <<-\\EOF &&\n+\t> Y\n+\tR X\n+\t< B\n+\tEOF\n+\ttest_cmp expect actual\n+'\n+\n+test_done\n"},{"id":"414573","messageId":"xmqq35yzknbr.fsf@gitster.c.googlers.com","threadId":"54990","inReplyTo":"xmqqsg6zkwa8.fsf@gitster.c.googlers.com","subject":"Re: [PATCH 2/2] revision: implement --show-linear-break for --graph","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2021-01-18T02:09:44Z","receivedAt":"2021-01-18T02:14:48Z","isPatch":true,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Junio C Hamano writes:\n\n> In other words, revs->break_revision_mark is left NULL unless\n> --show-linear-break is given.\n>\n>> @@ -4192,8 +4192,8 @@ const char *get_revision_mark(const struct rev_info *revs, const struct commit *\n>> \t\telse\n>> \t\t\treturn \">\";\n>> \t} else if (revs->graph) {\n>> -\t\tif (!commit->parents)\n>> -\t\t\treturn \"#\";\n>> +\t\tif (revs->break_revision_mark && !commit->parents)\n>> +\t\t\treturn revs->break_revision_mark;\n>\n> And that causes this to break. Now \"--graph\" alone won't show '#'\n> for the root commits, despite that is what [1/2] wanted to do.\n>\n> Here is a fix-up, plus some minimum tests. \n\nHaving said all that, I do not mind if the new markings were\nactivated only when --show-linear-break option (or a separate new\noption) is given. But if that is where we want to go, your [1/2]\nthat uses the new markings unconditionally is a regression.\n\nA better organization, if we wanted to have multiple and smaller\nsteps than a single whole thing, would be:\n\n [1/2] Introduce a new \"--mark-root-commits\" option, or abuse the\n existing \"--show-linear-break\" option, and change \"*<>\"\n marking used for commits to \"#LR\" (or whatever appropriate)\n when the option is in effect. Document the behaviour and add\n tests.\n\n [2/2] Introduce \"--show-linear-break=\" option.\n Document the behaviour and add tests.\n\nIf you apply [1/2] and [2/2] with the earlier fixes I sent, you'll\nsee many fallouts from existing tests, as the representation of the\nroot commit is changed unconditionally. We view breakages of tests\nas a rough estimate of how badly end-user scripts could break, and\nthe picture was not very pretty. And that is why I am suggesting\nthe above \"only do the new markings when asked, not unconditionally\"\napproach.\n\nI still am skeptical that spending 3 more letters to denote roots is\nworth it, though.\n\nThanks.\n"},{"id":"414585","messageId":"e0264a29-2112-f8c8-f066-2be445654d8e@pdinc.us","threadId":"54990","inReplyTo":"xmqq7dobmfrq.fsf@gitster.c.googlers.com","subject":"Re: [PATCH 1/2] revision: Denote root commits with '#'","fromName":"Kyle Marek","fromEmail":"kmarek@pdinc.us","sentAt":"2021-01-18T07:56:22Z","receivedAt":"2021-01-18T07:57:10Z","isPatch":true,"sender":{"key":"kmarek@pdinc.us","avatar":null},"body":"On 1/17/21 4:10 PM, Junio C Hamano wrote:\n> Kyle Marek writes:\n>\n>> This aids in identifying where an unrelated branch history starts when\n>> using `git log --graph --oneline --all`\n>>\n>> Signed-off-by: Kyle Marek\n>> ---\n>> revision.c | 6 ++++--\n>> 1 file changed, 4 insertions(+), 2 deletions(-)\n> No tests?\n\nI'm not very familiar with the code base. I now see the t/README file.\n\n>> diff --git a/revision.c b/revision.c\n>> index 9dff845bed..8556923de8 100644\n>> --- a/revision.c\n>> +++ b/revision.c\n>> @@ -4191,9 +4191,11 @@ const char *get_revision_mark(const struct rev_info *revs, const struct commit *\n>> \t\t\treturn \"<\";\n>> \t\telse\n>> \t\t\treturn \">\";\n>> -\t} else if (revs->graph)\n>> +\t} else if (revs->graph) {\n>> +\t\tif (!commit->parents)\n>> +\t\t\treturn \"#\";\n>> \t\treturn \"*\";\n>> -\telse if (revs->cherry_mark)\n>> +\t} else if (revs->cherry_mark)\n>> \t\treturn \"+\";\n>> \treturn \"\";\n>> }\n> Here is what I tried to come up with, but somehow the \"#\" marker is\n> not showing for me.\n>\n> The \"counted plus --left-right\" tests stress why a single \"#\" is not\n> good enough. I think the patch also needs to replace \"<\" and \">\"\n> for root commits that are left and right---in the tests, I used \"L\"\n> to denote \"root that is on the left side\" (and \"R\" for the right\n> side) instead of single \"#\", so that we do not to lose information.\n>\n> By the way, as I already said in the original thread, I do not think\n> the '#' marking is a good idea; I'd rather see the root commit shown\n> by shifting columns.\n\nSorry, I wasn't subscribed to the list until Jason CC'd me on his \nrequest. I also wasn't aware of --left-right.\n\nI'll investigate the revision-mark shifting idea. I am concerned that it \nwould get complicated if a graph edge extends around a revision that \nneeds to be shifted, but I'm finding it difficult to produce this with \n--graph:\n\n* 8d82d0a (HEAD -> master) Merge branch 'o1'\n|\\\n| * 3479914 (o1) O1\n| * a674e07 O1 <-- root commit\n| * 2237b52 (t) T\n| * f525fa5 T\n|/\n* f15f936 A\n| * 9e289ed (u) U\n|/\n* ee911c8 initial <-- root commit\n\nvs:\n\n* 8ee9b14 (HEAD -> master) Merge branch 'u'\n|\\\n| * ed1990f (u) U\n* | 277f31c Merge branch 'o1'\n|\\ \\\n| * | eaa71bb (o1) O1\n| * | 9203a43 O1 <-- root commit\n| /\n| | * bc2c4d9 (t) T\n| | * 2d3c03b T\n| |/\n|/|\n* | 6a26183 A\n|/\n* da85ccf initial <-- root commit\n \n\nThoughts? Will git ever graph something like:\n\n*\n|\\\n| *\n* |\n|\\ \\\n| * | <-- root commit\n| * | <-- some head\n|/ /\n* /\n|/\n* <-- root commit\n\n-- \n\n-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-\n- -\n- Kyle Marek PD Inc.http://www.pdinc.us -\n- Jr. Developer 10 West 24th Street #100 -\n- +1 (443) 269-1555 x361 Baltimore, Maryland 21218 -\n- -\n-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-\n\n"},{"id":"414586","messageId":"04c81462-3181-37d7-0109-4292040b84e9@pdinc.us","threadId":"54990","inReplyTo":"xmqq35yzknbr.fsf@gitster.c.googlers.com","subject":"Re: [PATCH 2/2] revision: implement --show-linear-break for --graph","fromName":"Kyle Marek","fromEmail":"kmarek@pdinc.us","sentAt":"2021-01-18T07:56:24Z","receivedAt":"2021-01-18T07:57:36Z","isPatch":true,"sender":{"key":"kmarek@pdinc.us","avatar":null},"body":"On 1/17/21 9:09 PM, Junio C Hamano wrote:\n> Junio C Hamano writes:\n>\n>> In other words, revs->break_revision_mark is left NULL unless\n>> --show-linear-break is given.\n>>\n>>> @@ -4192,8 +4192,8 @@ const char *get_revision_mark(const struct rev_info *revs, const struct commit *\n>>> \t\telse\n>>> \t\t\treturn \">\";\n>>> \t} else if (revs->graph) {\n>>> -\t\tif (!commit->parents)\n>>> -\t\t\treturn \"#\";\n>>> +\t\tif (revs->break_revision_mark && !commit->parents)\n>>> +\t\t\treturn revs->break_revision_mark;\n>> And that causes this to break. Now \"--graph\" alone won't show '#'\n>> for the root commits, despite that is what [1/2] wanted to do.\n>>\n>> Here is a fix-up, plus some minimum tests.\n> Having said all that, I do not mind if the new markings were\n> activated only when --show-linear-break option (or a separate new\n> option) is given. But if that is where we want to go, your [1/2]\n> that uses the new markings unconditionally is a regression.\n>\n> A better organization, if we wanted to have multiple and smaller\n> steps than a single whole thing, would be:\n>\n> [1/2] Introduce a new \"--mark-root-commits\" option, or abuse the\n> existing \"--show-linear-break\" option, and change \"*<>\"\n> marking used for commits to \"#LR\" (or whatever appropriate)\n> when the option is in effect. Document the behaviour and add\n> tests.\n>\n> [2/2] Introduce \"--show-linear-break=\" option.\n> Document the behaviour and add tests.\n>\n> If you apply [1/2] and [2/2] with the earlier fixes I sent, you'll\n> see many fallouts from existing tests, as the representation of the\n> root commit is changed unconditionally. We view breakages of tests\n> as a rough estimate of how badly end-user scripts could break, and\n> the picture was not very pretty. And that is why I am suggesting\n> the above \"only do the new markings when asked, not unconditionally\"\n> approach.\n\nSorry. I didn't make this clear. It is not an accident that patch 1 \ndenotes root commits unconditionally and patch 2 makes it optional. I \npresent two choices. If we prefer to unconditionally denote root \ncommits, patch 2 may be left out, otherwise, patch 1 should be squashed \naway.\n\nI didn't have an opinion towards either option, but you make a good \npoint about end-user scripts.\n\n> I still am skeptical that spending 3 more letters to denote roots is\n> worth it, though.\n\nMe too, but I think a user-defined mark needs to be a string to support \nUnicode characters.\n\n-- \n\n-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-\n- -\n- Kyle Marek PD Inc.http://www.pdinc.us -\n- Jr. Developer 10 West 24th Street #100 -\n- +1 (443) 269-1555 x361 Baltimore, Maryland 21218 -\n- -\n-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-\n\n"},{"id":"414613","messageId":"xmqqwnwajbuj.fsf@gitster.c.googlers.com","threadId":"54990","inReplyTo":"e0264a29-2112-f8c8-f066-2be445654d8e@pdinc.us","subject":"Re: [PATCH 1/2] revision: Denote root commits with '#'","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2021-01-18T19:15:16Z","receivedAt":"2021-01-18T19:52:59Z","isPatch":true,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Kyle Marek writes:\n\n> I'll investigate the revision-mark shifting idea. I am concerned that\n> it would get complicated if a graph edge extends around a revision\n> that needs to be shifted,...\n\nThe \"current graph layout makes it harder to see where the root is\"\nproblem has a natural solution: fix the graph layout so that the\nroot is easily visible.\n\nI however think it is a much harder approach to solve than using a\ndifferent mark for root commits, and it is the reason why there have\nbeen at least a few attempts in the past that did essentially the\nsame patch as yours, plus the \"linear break\" which we accepted.\n\n> * 8d82d0a (HEAD -> master) Merge branch 'o1'\n> |\\\n> | * 3479914 (o1) O1\n> | * a674e07 O1 <-- root commit\n> | * 2237b52 (t) T\n> | * f525fa5 T\n> |/\n> * f15f936 A\n> | * 9e289ed (u) U\n> |/\n> * ee911c8 initial <-- root commit\n>\n> vs:\n>\n> * 8ee9b14 (HEAD -> master) Merge branch 'u'\n> |\\\n> | * ed1990f (u) U\n> * | 277f31c Merge branch 'o1'\n> |\\ \\\n> | * | eaa71bb (o1) O1\n> | * | 9203a43 O1 <-- root commit\n> | /\n> | | * bc2c4d9 (t) T\n> | | * 2d3c03b T\n> | |/\n> |/|\n> * | 6a26183 A\n> |/\n> * da85ccf initial <-- root commit\n\nSorry, I am not quite sure what you are trying to illustrate with\nthe comparison between the above two. The latter makes it clear\nthat 9203a43 and da85ccf do not have parents in the depicted part of\nthe history [*1*].\n\nIn the former one, does 2237b52 have no child in the depicted part of\nthe history, and is the problem that it appears as if it has a674e07\nas a child? I wonder if we can just shift them, either:\n\n> * 8d82d0a (HEAD -> master) Merge branch 'o1'\n> |\\__\n> | * 3479914 (o1) O1\n> | * a674e07 O1 <-- root commit\n> | * 2237b52 (t) T\n> | * f525fa5 T\n> |/\n> * f15f936 A\n\nor\n\n> * 8d82d0a (HEAD -> master) Merge branch 'o1'\n> |\\\n> | * 3479914 (o1) O1\n> | * a674e07 O1 <-- root commit\n> | * 2237b52 (t) T\n> | __* f525fa5 T\n> |/\n> * f15f936 A\n\nOr we could punt to show it with an extra blank line, although it is\nsuboptimial.\n\n> * 8d82d0a (HEAD -> master) Merge branch 'o1'\n> |\\\n> | * 3479914 (o1) O1\n> | * a674e07 O1 <-- root commit\n> |\n> | * 2237b52 (t) T\n> | * f525fa5 T\n> |/\n> * f15f936 A\n\n\n[Footnote]\n\n*1* Stepping back a bit, I think concentrating too much on \"is it\n root?\" is a wrong way to think about the problem. Suppose you\n have two histories, e.g. (time flows from left to right; A and X\n are roots)\n\n A---B\n \\\n X---Y---Z\n\n and doing \"git log --graph --oneline Z\" would show A, B, X, Y\n and Z.\n\n If it benefits to show \"A\" (and \"X\") specially in the graph,\n that would mean that the current algorithm would show some other\n commit after showing A (probably X if it goes in chronological\n order), and it probably is confusing because X is shown on the\n same column as A, when there is no parent-child relationship\n between them (A is root after all).\n\n We are trying to highlight that A is not a child of anybody by\n using '#' instead.\n\n But in a slightly modified graph:\n\n C\n /\n O---A---B\n \\\n X---Y---Z\n\n if you do \"git log --graph --oneline C..Z\", you should see the\n same commits listed as above (A, B, X, Y and Z), and most likely\n in the same order.\n\n So special casing by \"Ah, A is a root commit, so let's show it\n with '#'\" does not help, even though we are facing exactly the\n same problem in the latter graph.\n\n And the right way to look at it is \"does A have any parent in\n the part of the history being shown?\", not \"does A have any\n parent?\" Then 'A' will get exactly the same treatment in the\n two examples, and the visual problem that makes A appear as if\n it has parent-child relationship with unrelated commit X goes\n away.\n\n \n"},{"id":"414618","messageId":"xmqqr1mij88k.fsf@gitster.c.googlers.com","threadId":"54990","inReplyTo":"xmqqwnwajbuj.fsf@gitster.c.googlers.com","subject":"Re: [PATCH 1/2] revision: Denote root commits with '#'","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2021-01-18T20:33:15Z","receivedAt":"2021-01-18T20:34:46Z","isPatch":true,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Junio C Hamano writes:\n\n> [Footnote]\n>\n> *1* Stepping back a bit, I think concentrating too much on \"is it\n> root?\" is a wrong way to think about the problem. Suppose you\n> have two histories, e.g. (time flows from left to right; A and X\n> are roots)\n\nA shorter and more concrete example. Start from an empty repository:\n\n\t$ git init\n\t$ git commit --allow-empty -m Aroot\n\t$ git checkout --orphan side\n\t$ git commit --allow-empty -m Xroot\n\t$ git log --all --graph --oneline\n * a1f7cb2 (HEAD -> side) Xroot\n * b6fb655 (master) Aroot\n\nThese depict two root commits, Aroot and Xroot, and no other\ncommits. We do want to show that these two commits do not have\nparent-child relationship at all, and your (and a few proposals made\nby other in the past) solution was to show them both with \"#\".\n\nContinuing in the same repository:\n\n\t$ git checkout --orphan another\n\t$ git commit --allow-empty -m Oroot\n\t$ git commit --allow-empty -m A\n\t$ git log --graph --oneline ^another^ another side\n * eddf116 (HEAD -> another) A\n * a1f7cb2 (side) Xroot\n\nThese depict two commits, A and Xroot, and no other commits. We\nalso want to show that these two commits do not have parent-child\nrelationship at all, but if we paint Xroot with \"#\", it still makes\nit appear that A is a child of Xroot.\n\n> And the right way to look at it is \"does A have any parent in\n> the part of the history being shown?\", not \"does A have any\n> parent?\" Then 'A' will get exactly the same treatment in the\n> two examples, and the visual problem that makes A appear as if\n> it has parent-child relationship with unrelated commit X goes\n> away.\n\nSo the condition we saw in your patches, !commit->parents, which\nattempted to see if it was root, needs to be replaced with a helper\nfunction that checks if there is any parent that is shown in the\noutput. Perhaps\n\n\tint no_interesting_parents(struct commit *commit)\n\t{\n\t\tstruct commit_list *parents = commit->parents;\n\n\t\twhile (parents) {\n\t\t\tif (!(parents->object.flags & UNINTERESTING))\n\t\t\t\treturn 0;\n\t\t\tparents = parents->next;\n\t\t}\n\t\treturn 1;\n\t}\n\nor something like that should serve as a replacement, i.e.\n\n\treturn !commit->parents ? \"#\" : \"*\";\n\nwould become\n\n\treturn no_interesting_parents(commit) ? \"#\" : \"*\";\n\nHmm?\n\n"},{"id":"414629","messageId":"xmqqmtx6j6wt.fsf@gitster.c.googlers.com","threadId":"54990","inReplyTo":"04c81462-3181-37d7-0109-4292040b84e9@pdinc.us","subject":"Re: [PATCH 2/2] revision: implement --show-linear-break for --graph","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2021-01-18T21:01:54Z","receivedAt":"2021-01-18T21:10:37Z","isPatch":true,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Kyle Marek writes:\n\n> Me too, but I think a user-defined mark needs to be a string to\n> support Unicode characters.\n\nAhh, I didn't even consider making it user-defined.\n\nAs it seems a lot safer to make this an optional feature, it does\nsort-of make sense to let the letters used for root & left-root be\ncustomizable, and it does make sense to take a multi-byte character,\nbut I am not sure what implications it has if we allowed any string\nwithout ensuring that it occupies one display column.\n\n"},{"id":"414669","messageId":"04380a95-b8cf-f246-e496-dc469f617eb5@pdinc.us","threadId":"54990","inReplyTo":"xmqqmtx6j6wt.fsf@gitster.c.googlers.com","subject":"Re: [PATCH 2/2] revision: implement --show-linear-break for --graph","fromName":"Kyle Marek","fromEmail":"kmarek@pdinc.us","sentAt":"2021-01-19T07:44:02Z","receivedAt":"2021-01-19T07:48:18Z","isPatch":true,"sender":{"key":"kmarek@pdinc.us","avatar":null},"body":"On 1/18/21 4:01 PM, Junio C Hamano wrote:\n> Kyle Marek writes:\n>\n>> Me too, but I think a user-defined mark needs to be a string to\n>> support Unicode characters.\n> Ahh, I didn't even consider making it user-defined.\n>\n> As it seems a lot safer to make this an optional feature, it does\n> sort-of make sense to let the letters used for root & left-root be\n> customizable, and it does make sense to take a multi-byte character,\n> but I am not sure what implications it has if we allowed any string\n> without ensuring that it occupies one display column.\n\nDoes git, or a dependency library, have the ability to interpret TERM \nand locale to determine on-screen character count/size?\n\nIf not, maybe let users use multi-character strings, but call it misuse \nof the option that will mess offset that row of the --graph output until \nwe have something to determine on-screen size.\n\n-- \n\n-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-\n- -\n- Kyle Marek PD Inc. http://www.pdinc.us -\n- Jr. Developer 10 West 24th Street #100 -\n- +1 (443) 269-1555 x361 Baltimore, Maryland 21218 -\n- -\n-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-\n\n"},{"id":"414672","messageId":"237aeef3-239f-bff4-fa17-5581092c8f51@pdinc.us","threadId":"54990","inReplyTo":"xmqqr1mij88k.fsf@gitster.c.googlers.com","subject":"Re: [PATCH 1/2] revision: Denote root commits with '#'","fromName":"Kyle Marek","fromEmail":"kmarek@pdinc.us","sentAt":"2021-01-19T07:43:57Z","receivedAt":"2021-01-19T07:56:07Z","isPatch":true,"sender":{"key":"kmarek@pdinc.us","avatar":null},"body":"On 1/18/21 3:33 PM, Junio C Hamano wrote:\n> Junio C Hamano writes:\n>\n>> [Footnote]\n>>\n>> *1* Stepping back a bit, I think concentrating too much on \"is it\n>> root?\" is a wrong way to think about the problem. Suppose you\n>> have two histories, e.g. (time flows from left to right; A and X\n>> are roots)\n> A shorter and more concrete example. Start from an empty repository:\n>\n> \t$ git init\n> \t$ git commit --allow-empty -m Aroot\n> \t$ git checkout --orphan side\n> \t$ git commit --allow-empty -m Xroot\n> \t$ git log --all --graph --oneline\n> * a1f7cb2 (HEAD -> side) Xroot\n> * b6fb655 (master) Aroot\n>\n> These depict two root commits, Aroot and Xroot, and no other\n> commits. We do want to show that these two commits do not have\n> parent-child relationship at all, and your (and a few proposals made\n> by other in the past) solution was to show them both with \"#\".\n>\n> Continuing in the same repository:\n>\n> \t$ git checkout --orphan another\n> \t$ git commit --allow-empty -m Oroot\n> \t$ git commit --allow-empty -m A\n> \t$ git log --graph --oneline ^another^ another side\n> * eddf116 (HEAD -> another) A\n> * a1f7cb2 (side) Xroot\n>\n> These depict two commits, A and Xroot, and no other commits. We\n> also want to show that these two commits do not have parent-child\n> relationship at all, but if we paint Xroot with \"#\", it still makes\n> it appear that A is a child of Xroot.\n>\n>> And the right way to look at it is \"does A have any parent in\n>> the part of the history being shown?\", not \"does A have any\n>> parent?\" Then 'A' will get exactly the same treatment in the\n>> two examples, and the visual problem that makes A appear as if\n>> it has parent-child relationship with unrelated commit X goes\n>> away.\n> So the condition we saw in your patches, !commit->parents, which\n> attempted to see if it was root, needs to be replaced with a helper\n> function that checks if there is any parent that is shown in the\n> output. Perhaps\n>\n> \tint no_interesting_parents(struct commit *commit)\n> \t{\n> \t\tstruct commit_list *parents = commit->parents;\n>\n> \t\twhile (parents) {\n> \t\t\tif (!(parents->object.flags & UNINTERESTING))\n> \t\t\t\treturn 0;\n> \t\t\tparents = parents->next;\n> \t\t}\n> \t\treturn 1;\n> \t}\n>\n> or something like that should serve as a replacement, i.e.\n>\n> \treturn !commit->parents ? \"#\" : \"*\";\n>\n> would become\n>\n> \treturn no_interesting_parents(commit) ? \"#\" : \"*\";\n>\n> Hmm?\n\nOkay, I see what you mean. Fixing --graph to avoid implying ancestry \nsounds like a better approach to me.\n\nThat being said, I spoke to Jason recently, and he expressed interest in \noptionally marking root commits so they are easy to search for in a \ngraph with something like /# in `less`. I see value in this, too.\n\nSo would you be open to my modifying of the patch in question (patch 1+2 \nsquashed, I guess) to instead use \"--mark-roots=\" to optionally \nmark root commits with a string , and pursue fixing the --graph \nrendering issue in another series?\n\nIf so, what would you like to see out of the --left-right issue? Maybe \n\"--mark-left-root=\" and \"--mark-right-root=\", so multi-byte \nstrings may be used? Can there be more than one root on either side? (so \nthe option would use a plural \"roots\" instead of \"root\"?)\n\n-- \n\n-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-\n- -\n- Kyle Marek PD Inc. http://www.pdinc.us -\n- Jr. Developer 10 West 24th Street #100 -\n- +1 (443) 269-1555 x361 Baltimore, Maryland 21218 -\n- -\n-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-\n\n"},{"id":"414733","messageId":"xmqq1reginnq.fsf@gitster.c.googlers.com","threadId":"54990","inReplyTo":"237aeef3-239f-bff4-fa17-5581092c8f51@pdinc.us","subject":"Re: [PATCH 1/2] revision: Denote root commits with '#'","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2021-01-19T22:10:01Z","receivedAt":"2021-01-19T22:11:27Z","isPatch":true,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Kyle Marek writes:\n\n>> So the condition we saw in your patches, !commit->parents, which\n>> attempted to see if it was root, needs to be replaced with a helper\n>> function that checks if there is any parent that is shown in the\n>> output.\n>> ...\n>> Hmm?\n>\n> Okay, I see what you mean. Fixing --graph to avoid implying ancestry\n> sounds like a better approach to me.\n\nSorry, I do not know how you drew that conclusion from my\ndescription.\n\nAll I meant to convey is \"roots are not special at all, commits that\ndo not have parents in the parts of the history shown are, and care\nmust be taken to ensure that they do not appear to have parents\".\n\nAnd the argument applies equally to either of two approaches.\nWhether the solution chosen is\n\n (1) to use special set of markers \"{#}\" for commits that do not\n have parents in the displayed part of the history instead of\n the usual \"<*>\", or\n\n (2) to stick to the normal set of markers \"<*>\" but shift the graph\n to avoid false ancestry.\n\nwe shouldn't be special casing \"root commits\" just because they are\nroots. Exactly the same issue exists for non-root commits whose\nparents are not shown in the output, if commits from unrelated\nancestry is drawn directly below them.\n\n> That being said, I spoke to Jason recently, and he expressed interest\n> in optionally marking root commits so they are easy to search for in a \n> graph with something like /# in `less`. I see value in this,\n\nI do not mind to denote the \"this commit may appear directly on top\nof another commit, but there is no ancestry\" situation with a\nspecial set of markers that is different from the usual \"<*>\" (for\nleft, normal and right) set. I agree pagers are good ways to /search\nthings in the output.\n\n> So would you be open to my modifying of the patch in question (patch\n> 1+2 squashed, I guess) to instead use \"--mark-roots=\" to\n> optionally mark root commits with a string , and pursue fixing\n> the --graph rendering issue in another series?\n\nI do not mind if the graph rendering fix does not happen yet again;\nIIRC the past contributors couldn't implement it, either.\n\nI think this new feature should be made opt-in by introducing a new\noption (without giving it a configuration variable), with explicit\n\"--no-