RFC post-rewrite / automatic notes copying
Get series with b4
b4 am [email protected]Versions newest first · linked by author and title
- v2 RFC Latest11 patches 9 repliesIn review
- v1 RFC6 patches 2 repliesSuperseded
Patches in v2 No applies recorded yet
| # | Subject and files | Size | Replies | Applied |
|---|---|---|---|---|
| 1/11 | Documentation: document post-rewrite hookDocumentation/ | +36 | 3 replies | |
| 2/11 | commit --amend: invoke post-rewrite hookbuiltin-commit.c git-rebase--interactive.sh t/ | +96 | 0 replies | |
| 3/11 | rebase: invoke post-rewrite hookgit-am.sh git-rebase.sh t/ | +45 | 0 replies | |
| 4/11 | rebase -i: invoke post-rewrite hookgit-rebase--interactive.sh t/ | +117 | 0 replies | |
| 5/11 | notes: clean up t3301t/ | +46 | 1 reply | |
| 6/11 | notes: implement 'git notes copy --stdin'Documentation/ builtin-notes.c notes.c +2 more | +129 | 0 replies | |
| 7/11 | notes: implement helpers needed for note copying during rewriteDocumentation/ Documentation/ Documentation/ +3 more | +219 | 5 replies | |
| 8/11 | rebase: support automatic notes copyinggit-am.sh git-rebase--interactive.sh git-rebase.sh +2 more | +35 | 0 replies | |
| 9/11 | commit --amend: copy notes to the new commitbuiltin-commit.c t/ | +17 | 0 replies | |
| 10/11 | filter-branch: invoke post-rewrite hookDocumentation/ git-filter-branch.sh t/ | +55 | 0 replies | |
| 11/11 | filter-branch: learn how to filter notesDocumentation/ Documentation/ Documentation/ +2 more | +64 | 2 replies |
Reviews trailers per patch
None of the 11 patches has a Reviewed-by or Acked-by of their own; the series has one for all of them.
| Patch | Johan Herland |
|---|---|
| Series | A-b A-b A-b A-b A-b A-b A-b A-b A-b A-b A-b A-b A-b A-b |
| 1/11 | |
| 2/11 | |
| 3/11 | |
| 4/11 | |
| 5/11 | |
| 6/11 | |
| 7/11 | |
| 8/11 | |
| 9/11 | |
| 10/11 | |
| 11/11 |
Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes.
Cover letter [RFC PATCH v2 00/11] post-rewrite / automatic notes copying
I originally had two series, but to retain my sanity while shuffling fixups, I based the entire batch on a rebase of jh/notes onto current master. The first four do not depend on jh/notes though, and are almost unchanged from the first round.
The docs are probably sprinkled a bit strangely over the patches, not sure if it's worth a cleanup though (or I could just gather all the docs in a final patch). At least tests should be in the right place.
Based on Hannes's comment:
Quoted 10 lines · On Monday 15 February 2010 21:36:05 Johannes Sixt wrote:
Review
Quoted 4 lines · Thomas Rast
(quoted diff) +4
-0+if test -x "$GIT_DIR"/hooks/post-rewrite && + test -s "$workdir"/../rewritten; then + "$GIT_DIR"/hooks/post-rewrite filter-branch < "$workdir"/../rewritten +fiThis sounds extra-strange. As if filter-branch is used 20 times a day. If the intent is to carry notes over to new commits, then filter-branch should grow a --notes-filter instruction, no? With an easily accessible copy-everything mode like --tagname-filter=cat.
I made filter-branch extra-tweakable, but tried to stay consistent with the configuration of the rest of the series. Maybe it's a bit overengineered now, dunno.
Quoted 6 lines · On Monday 15 February 2010 02:25:52 Johan Herland wrote:
Quoted 5 lines · Thomas Rast
I spent some time trying to refactor
cmd_notes()into something that can nicely handle commands that do not fit the normal scheme, but eventually was too tired to continue and just crafted it in the existing code in the right place.
[...]
Quoted 2 lines
As you say,
cmd_notes()should be refactored, so that you don't have to duplicate all of this (and some of the below stuff).
Actually I meant something different: it took me quite some time to figure out what goes where in the flow of the current main() function, since it is sprinkled with if(subcommand) -- for most of them, several such clauses.
I guess it made sense at the time, because all commands more or less worked with the same arguments and mostly the same background logic, just in a slightly different way. But I think it's quite unnatural for an entirely-different command like 'copy --stdin'.
But maybe that's just me.
Quoted 5 lines
We (this goes for the existing
commit_notes()call incmd_notes()as well) should either check the return value forcommit_notes(), or (since it currently only returns 0) we should changecommit_notes()to void instead of int. Actually, the correct solution is probably to "returnerror()" instead of "die()" insidecommit_notes(), and then check the return value.
I'll leave that up to your judgement; I haven't looked into the API workings much, so I'm not sure when it should fail how.
[Thanks for the other comments on my embarrassing mistakes; I was apparently rather too tired to continue programming.]
Thomas Rast (11):
Documentation: document post-rewrite hook
commit --amend: invoke post-rewrite hook
rebase: invoke post-rewrite hook
rebase -i: invoke post-rewrite hook
notes: clean up t3301
notes: implement 'git notes copy --stdin'
notes: implement helpers needed for note copying during rewrite
rebase: support automatic notes copying
commit --amend: copy notes to the new commit
filter-branch: invoke post-rewrite hook
filter-branch: learn how to filter notes
Documentation/config.txt | 17 +++ Documentation/git-filter-branch.txt | 20 +++- Documentation/git-notes.txt | 16 ++- Documentation/githooks.txt | 40 +++++ builtin-commit.c | 49 +++++++ builtin-notes.c | 134 +++++++++++++++++- builtin.h | 15 ++ git-am.sh | 13 ++ git-filter-branch.sh | 37 +++++ git-rebase--interactive.sh | 45 ++++++- git-rebase.sh | 6 + notes.c | 21 +++ notes.h | 9 ++ t/t3301-notes.sh | 273 +++++++++++++++++++++++------------ t/t3400-rebase.sh | 16 ++ t/t3404-rebase-interactive.sh | 10 ++ t/t5407-post-rewrite-hook.sh | 188 ++++++++++++++++++++++++ t/t7003-filter-branch.sh | 22 +++ t/t7501-commit.sh | 11 ++ 19 files changed, 843 insertions(+), 99 deletions(-) create mode 100755 t/t5407-post-rewrite-hook.sh