Re: [PATCH] worktree: allow "-" short-hand for @{-1} in add command
Jordan DE GEAwrites: >>> +test_expect_success '"add" using shorthand - fails when no previous >>> branch' ' >>> + test_must_fail git worktree add existing - >>> +' >> >> Just an observation, but the error message we would see here might >> be interesting. > > Of course, that’s useful to be sure of the error, I will do in next preroll. That was not what I meant. The exit status being non-zero is what we primarily care about. The exact phrasing of the error message is much less important and in general we shouldn't enforce "the error message must remain so" in the test. If you observe the error message from this test, e.g. by running it with "-v", I suspect that you would see the message would complain about "@{-1}". I just wanted to make sure that you saw it and thought about its ramifications. It is perfectly fine by me (others might disagree, though) if your conclusion after thinking about it is "Even though the user may be surprised to get complaints for "@{-1}" that she never gave to the command (she gave "-"), because we clearly document that "-" is a mere synonym/short-hand for @{-1}, it is OK". I still want to see that behaviour justified in the proposed commit log message. And that is why I said it "might be interesting". -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH] worktree: allow "-" short-hand for @{-1} in add command
Matthieu Moywrites: > Junio C Hamano writes: > >> Jordan DE GEA writes: >> >>> + branch=$(cd short-hand && git rev-parse --symbolic-full-name HEAD) && >>> + test "$branch" = refs/heads/newbranch && >>> + cd .. >> >> If any of the command between "cd short-hand" and "cd .." failed, >> after correcting the broken &&-chain, the next test will end up >> running in short-hand directory, which it is not expecting. A >> canonical way to avoid this problem is to replace the above with: >> >> ... >> git worktree add short-hand - && >> ( >> cd short-hand && >> ... >> test "$branch" = refs/heads/newbranch >> ) > > Actually, $(...) implicitly does a subshell, so the "cd .." was just > useless. You trimmed my message a bit too aggressively while composing your response, and I think that is what ended up confusing you. Here is what I wrote: | > + cd short-hand && | > + test $(git rev-parse --symbolic-full-name HEAD) = "refs/heads/newbranch" | | Broken &&-chain. | | > + branch=$(cd short-hand && git rev-parse --symbolic-full-name HEAD) && | > + test "$branch" = refs/heads/newbranch && | > + cd .. | | If any of the command between "cd short-hand" and "cd .." failed, | ... The problematic "cd short-hand" is the one a few lines above where you started quoting, not the one you saw in the $(). -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH] worktree: allow "-" short-hand for @{-1} in add command
>> +test_expect_success '"add" using shorthand - fails when no previous branch' >> ' >> +test_must_fail git worktree add existing - >> +' > > Just an observation, but the error message we would see here might > be interesting. Of course, that’s useful to be sure of the error, I will do in next preroll. > >> +branch=$(cd short-hand && git rev-parse --symbolic-full-name HEAD) && >> +test "$branch" = refs/heads/newbranch && >> +cd .. > > If any of the command between "cd short-hand" and "cd .." failed, > after correcting the broken &&-chain, the next test will end up > running in short-hand directory, which it is not expecting. A > canonical way to avoid this problem is to replace the above with: > > ... >git worktree add short-hand - && >( > cd short-hand && >... >test "$branch" = refs/heads/newbranch > ) > > In this particular case, alternatively, you could also do something > like this: > >git worktree add short-hand - && > echo refs/heads/newbranch >expect && > git -C short-hand rev-parse --symbolic-full-name HEAD >actual && > test_cmp expect actual Yes, that’s a good idea. I take these lines. -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH] worktree: allow "-" short-hand for @{-1} in add command
Junio C Hamanowrites: > Jordan DE GEA writes: > >> +branch=$(cd short-hand && git rev-parse --symbolic-full-name HEAD) && >> +test "$branch" = refs/heads/newbranch && >> +cd .. > > If any of the command between "cd short-hand" and "cd .." failed, > after correcting the broken &&-chain, the next test will end up > running in short-hand directory, which it is not expecting. A > canonical way to avoid this problem is to replace the above with: > > ... > git worktree add short-hand - && > ( > cd short-hand && > ... > test "$branch" = refs/heads/newbranch > ) Actually, $(...) implicitly does a subshell, so the "cd .." was just useless. > git -C short-hand rev-parse --symbolic-full-name HEAD >actual && Indeed, git -C is an even better way to say "cd .. && git ..." -- Matthieu Moy http://www-verimag.imag.fr/~moy/ -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH] worktree: allow "-" short-hand for @{-1} in add command
Jordan DE GEAwrites: > From: Jordan DE GEA > > Since `git worktree add` uses `git checkout` when `[]` is used, > and `git checkout -` is already supported, it makes sense to allow the > same shortcut in `git worktree add`. OK. > > Signed-off-by: Matthieu Moy > Signed-off-by: Jordan DE GEA > --- > Documentation/git-worktree.txt | 3 ++- > builtin/worktree.c | 3 +++ > t/t2025-worktree-add.sh| 18 ++ > 3 files changed, 23 insertions(+), 1 deletion(-) > > diff --git a/Documentation/git-worktree.txt b/Documentation/git-worktree.txt > index c622345..48e5fdf 100644 > --- a/Documentation/git-worktree.txt > +++ b/Documentation/git-worktree.txt > @@ -48,7 +48,8 @@ add []:: > > Create `` and checkout `` into it. The new working directory > is linked to the current repository, sharing everything except working > -directory specific files such as HEAD, index, etc. > +directory specific files such as HEAD, index, etc. The last branch you > +were on can be specify with `-` as `` which is synonymous with > `"@{-1}"`. You meant "can be specified", I think. Fixing it would make the line a bit too long, so fold it around the word "synonymous". > diff --git a/t/t2025-worktree-add.sh b/t/t2025-worktree-add.sh > index 3acb992..b713efb 100755 > --- a/t/t2025-worktree-add.sh > +++ b/t/t2025-worktree-add.sh > @@ -18,6 +18,24 @@ test_expect_success '"add" an existing empty worktree' ' > git worktree add --detach existing_empty master > ' > > +test_expect_success '"add" using shorthand - fails when no previous branch' ' > + test_must_fail git worktree add existing - > +' Just an observation, but the error message we would see here might be interesting. > +test_expect_success '"add" using - shorthand' ' > + git checkout -b newbranch && > + echo hello >myworld && > + git add myworld && > + git commit -m myworld && > + git checkout master && > + git worktree add short-hand - && > + cd short-hand && > + test $(git rev-parse --symbolic-full-name HEAD) = "refs/heads/newbranch" Broken &&-chain. > + branch=$(cd short-hand && git rev-parse --symbolic-full-name HEAD) && > + test "$branch" = refs/heads/newbranch && > + cd .. If any of the command between "cd short-hand" and "cd .." failed, after correcting the broken &&-chain, the next test will end up running in short-hand directory, which it is not expecting. A canonical way to avoid this problem is to replace the above with: ... git worktree add short-hand - && ( cd short-hand && ... test "$branch" = refs/heads/newbranch ) In this particular case, alternatively, you could also do something like this: git worktree add short-hand - && echo refs/heads/newbranch >expect && git -C short-hand rev-parse --symbolic-full-name HEAD >actual && test_cmp expect actual and it would be sufficient. It is not immediately obvious to me why you have two copies of the same test in your patch to see where HEAD points at. If the reason is because you suspect that "git -C $there" form may give subtly different behaviour and wanted to test both, then you could do something like this: git worktree add short-hand - && echo refs/heads/newbranch >expect && git -C short-hand rev-parse --symbolic-full-name HEAD >actual && test_cmp expect actual && (cd short-hand && git rev-parse --symbolic-full-name HEAD) >actual && test_cmp expect actual but I do not think that is necessary. This test is not about "does rev-parse --symbolic-full-name work correctly with 'git -C $there'?" Thanks. -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
[PATCH] worktree: allow "-" short-hand for @{-1} in add command
From: Jordan DE GEASince `git worktree add` uses `git checkout` when `[]` is used, and `git checkout -` is already supported, it makes sense to allow the same shortcut in `git worktree add`. Signed-off-by: Matthieu Moy Signed-off-by: Jordan DE GEA --- Documentation/git-worktree.txt | 3 ++- builtin/worktree.c | 3 +++ t/t2025-worktree-add.sh| 18 ++ 3 files changed, 23 insertions(+), 1 deletion(-) diff --git a/Documentation/git-worktree.txt b/Documentation/git-worktree.txt index c622345..48e5fdf 100644 --- a/Documentation/git-worktree.txt +++ b/Documentation/git-worktree.txt @@ -48,7 +48,8 @@ add []:: Create `` and checkout `` into it. The new working directory is linked to the current repository, sharing everything except working -directory specific files such as HEAD, index, etc. +directory specific files such as HEAD, index, etc. The last branch you +were on can be specify with `-` as `` which is synonymous with `"@{-1}"`. + If `` is omitted and neither `-b` nor `-B` nor `--detached` used, then, as a convenience, a new branch based at HEAD is created automatically, diff --git a/builtin/worktree.c b/builtin/worktree.c index d8e3795..d800d47 100644 --- a/builtin/worktree.c +++ b/builtin/worktree.c @@ -340,6 +340,9 @@ static int add(int ac, const char **av, const char *prefix) path = prefix ? prefix_filename(prefix, strlen(prefix), av[0]) : av[0]; branch = ac < 2 ? "HEAD" : av[1]; + if (!strcmp(branch, "-")) + branch = "@{-1}"; + opts.force_new_branch = !!new_branch_force; if (opts.force_new_branch) { struct strbuf symref = STRBUF_INIT; diff --git a/t/t2025-worktree-add.sh b/t/t2025-worktree-add.sh index 3acb992..b713efb 100755 --- a/t/t2025-worktree-add.sh +++ b/t/t2025-worktree-add.sh @@ -18,6 +18,24 @@ test_expect_success '"add" an existing empty worktree' ' git worktree add --detach existing_empty master ' +test_expect_success '"add" using shorthand - fails when no previous branch' ' + test_must_fail git worktree add existing - +' + +test_expect_success '"add" using - shorthand' ' + git checkout -b newbranch && + echo hello >myworld && + git add myworld && + git commit -m myworld && + git checkout master && + git worktree add short-hand - && + cd short-hand && + test $(git rev-parse --symbolic-full-name HEAD) = "refs/heads/newbranch" + branch=$(cd short-hand && git rev-parse --symbolic-full-name HEAD) && + test "$branch" = refs/heads/newbranch && + cd .. +' + test_expect_success '"add" refuses to checkout locked branch' ' test_must_fail git worktree add zere master && ! test -d zere && -- 2.7.4 (Apple Git-66) -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html