3 this is a temporary file for the seq-builtin-dev (and thus seq-builtin-rfc)
4 branch that should collect information of what is still to do in the
5 sequencer builtin to be "ready" to be sent to the list.
6 (Remark: I personally use sequencer since GSoC 08 mid-time, so it is usable,
7 but it's not as perfect as it could be.)
9 The problem is that I have far too little time for further Git sequencer
10 development, so I thought it is helpful (for interested Git developers
11 or users) to make a file like this accessible in the repo.
15 1) TODO information from test suite:
18 Running t3404-rebase-interactive.sh
19 * still broken 32: avoid unnecessary reset
22 * still broken 20: am --committer-date-is-author-date
23 * still broken 22: am --ignore-date
24 * still broken 23: am into an unborn branch
28 2) TODO or TODO-like information from recent commit messages:
31 commit 0f6d62165f33012887d5874873a497f1769413a6
32 Merge: e7b8dab... e658002...
33 Author: Stephan Beyer <s-beyer@gmx.net>
34 Date: Mon Aug 3 03:55:27 2009 +0200
36 Merge branch 'junio/master' into seq-builtin-dev
38 This merge is a lazy merge that drops
40 1) git-am.sh changes introduced by:
41 ea10b60c91 Work around ash "alternate value" expansion bug
42 47c9739e5e am: simplify "sq" function by using "git rev-parse --sq-quote"
43 a5a6755a1d git-am foreign patch support: introduce patch_format
44 15ced753ac git-am foreign patch support: autodetect some patch formats
45 c574e683b5 git-am foreign patch support: StGIT support
46 0cd29a0371 git-am: refactor 'cleaning up and aborting'
48 2) git-rebase--interactive.sh changes introduced by:
50 1830d9cb62 Fix rebase -p --onto
52 and this merge also does not
54 3) introduce the new OPT_FILENAME option to builtin-sequencer.
56 commit e7b8dab0c2a73ade92017a52bb1405ea1534ef20
57 Merge: 0c73ae7... 99ddd24...
58 Author: Stephan Beyer <s-beyer@gmx.net>
59 Date: Wed May 20 10:54:37 2009 +0200
61 Merge branch 'junio/master' into seq-builtin-dev
63 After this merge f79d4c8 "teach git-am to apply a patch to an unborn
64 branch" has to be reimplemented in sequencer by allowing the "patch"
65 insn on unborn branches.
66 The related test in t/t4150-am.sh is set to test_expect_failure.
71 commit 0c73ae7c5ddfb348174fd6b85bf33f1a6b44c1ca
72 Merge: af91a89... e37347b...
73 Author: Stephan Beyer <s-beyer@gmx.net>
74 Date: Thu Apr 9 16:58:24 2009 +0200
76 Merge branch 'junio/master' into seq-builtin-dev
81 git-rebase--interactive.sh
83 0e757e30c726d9d8ae82bd9989be3cff5d230288 has to be applied to
84 builtin-sequencer.c to make it avoid unnecessary resets again
87 commit 71977c161725bbbea74b2b3fc24ec7756ceaf0a1
88 Merge: fd06e4e... 621f1b4...
89 Author: Stephan Beyer <s-beyer@gmx.net>
90 Date: Sun Feb 8 03:26:06 2009 +0100
92 Merge branch 'junio/master' into seq-builtin-dev
94 This merge leaves two test cases of t/t4150-am.sh with
95 expected failure, because the tested features are not yet
96 implemented into git sequencer.
100 git-rebase--interactive.sh
102 t/t3411-rebase-preserve-around-merges.sh
104 commit 4e7ab1d123550dc16799b40bbe101a5ccb120178
105 Author: Stephan Beyer <s-beyer@gmx.net>
106 Date: Sun Jan 18 22:56:48 2009 +0100
108 git-am: work around "Sequencing (./.)" messages by sequencer
110 We do not want "Sequencing (../..)" messages by sequencer, because we
111 have our own status messages.
113 This is only a workaround.
115 The best way is perhaps an option to sequencer telling what status
118 commit dea5325d61432a83688a92a5c35213456c548c95
119 Author: Stephan Beyer <s-beyer@gmx.net>
120 Date: Sun Jan 18 21:33:36 2009 +0100
122 sequencer: remove "Conflicts:" lines when picking (without -R)
124 This mimics the old git rebase -i behavior.
126 A better way could be to keep this, but run the prepare-commit-msg
127 hook with "merge" instead of "message" on --continue.
131 3) unsorted and unverified TODO information:
134 A long time ago I collected this information in an outliner tool with items
135 usually written down in German.
136 A pure dump of this information (German parts translated into English)
137 follows, but I cannot tell how much of it is useful or still true at all:
139 * 0e757e30c726d9d8ae82bd9989be3cff5d230288, see t3404 fail
141 * git-am --ignore-date and --commit-date-as-author-date
143 * git am: apply a patch to unborn branch: f79d4c8a0f22d7fd25018be846c7e
145 * sequencer insn counting: save "cur" and "max" and count correctly after
146 --continue (i.e. go on and not start at 1 again)
148 * Bug: see Sitaram Chamarty mail -> nice test case, fails currently
150 * Bug? see Constantine plot
152 * pick -r A ; squash A -> empty commit! Should _fail_ instead
154 * 4fb1a19d502f4a2ca47bfb75cce605f3447819e6 -> make_patch
155 * still to do: prettify diff -cc [see stash on host "littlefarm"]
156 * pick seq-builtin-rfc~5 printed conflict patch instead of
157 writing it to git-dir/sequencer/patch - why?
158 [seq-builtin-rfc~5 = 58522f29193f1fa6a8974250105053fbb46ccbec]
160 * check original vs mine: rebase --continue and one cherry-pick
161 -> does this insert "Conflicts:" in commit message?
162 * mine: rerere() is called AFTER printing "Automatic cherry-pick failed"
164 * on conflict: commit msg doesn't look like conflict after "git commit"
166 * check: c767184d8f99404d456b32a2c6c5c41d6db7a847 -> same behavior currently?
168 * check: does original rebase -i -p redo the changes of a _merge commit_?
169 If no, is this a feature or a bug and do we want it (after the
170 transition to sequencer without that feature)?
172 * case WHY_CONFLICT: why = WHY_UNKNOWN; in sequencer_continue?
173 Is this correct/useful?
175 * git-am/git-rebase -> new option for applying/rebasing status messages
177 * git-am: Dirty index: cannot apply patches (dirty: $files)
179 * Bug? git-rebase does not like fast-forward
181 * split rebase-i migration in rebase -i and rebase -i -p migration
182 into separate patches, so that rebase -i can be accepted when
183 rebase -i -p migration is not loved by everyone
185 * rebase -i -p is a topic on its own
187 * Dscho didn't like the way Junio had proposed it: using
188 "mark", "reset", "merge" insns
190 * after some revealed rebase -i -p foibles,
191 Dscho started a rewrite of rebase -i -p
194 http://repo.or.cz/w/git/dscho.git?a=blob_plain;hb=blog;f=index.html
198 * WHY_UNKNOWN output?
200 * improve reset_index_file()
201 - do we really need to _write_?
203 * squashing into a merge?
205 * fflush(stdout) on "Sequencing"?
209 * get rid of STDERR in merge
210 (Automatic merge went well; stopped before committing as requested)
212 * look, if commit_tree can be used in a useful way (inside do_commit())
216 * respect commit.signoff?
220 Stephan Beyer <s-beyer AT gmx.net>