* Re: [PATCH 2/2] commit: add --skip-intent-to-add to allow commit with i-t-a entries in index
From: Junio C Hamano @ 2012-01-11 23:55 UTC (permalink / raw)
To: Nguyễn Thái Ngọc Duy; +Cc: git
In-Reply-To: <1326275982-29866-3-git-send-email-pclouds@gmail.com>
Nguyễn Thái Ngọc Duy <pclouds@gmail.com> writes:
> ---
> builtin/commit.c | 10 +++++++---
> cache-tree.c | 8 +++++---
> cache-tree.h | 1 +
> t/t2203-add-intent.sh | 17 +++++++++++++++++
> 4 files changed, 30 insertions(+), 6 deletions(-)
>
> diff --git a/builtin/commit.c b/builtin/commit.c
> index bf42bb3..021206e 100644
> --- a/builtin/commit.c
> +++ b/builtin/commit.c
> @@ -86,6 +86,7 @@ static int all, also, interactive, patch_interactive, only, amend, signoff;
> static int edit_flag = -1; /* unspecified */
> static int quiet, verbose, no_verify, allow_empty, dry_run, renew_authorship;
> static int no_post_rewrite, allow_empty_message;
> +static int cache_tree_flags, skip_intent_to_add;
> static char *untracked_files_arg, *force_date, *ignore_submodule_arg;
> static char *sign_commit;
>
> @@ -170,6 +171,7 @@ static struct option builtin_commit_options[] = {
> OPT_BOOLEAN(0, "amend", &amend, "amend previous commit"),
> OPT_BOOLEAN(0, "no-post-rewrite", &no_post_rewrite, "bypass post-rewrite hook"),
> { OPTION_STRING, 'u', "untracked-files", &untracked_files_arg, "mode", "show untracked files, optional modes: all, normal, no. (Default: all)", PARSE_OPT_OPTARG, NULL, (intptr_t)"all" },
> + OPT_BOOL(0, "skip-intent-to-add", &skip_intent_to_add, "allow intent-to-add entries in index"),
This is more like "ignore", not "allow", from end user's point of view,
no? The user earlier said "I cannot decide what contents to put in the
commit yet for this path", and normally we catch it and remind the user
that she needs to decide. This option gives her a quick way to say "I
decide that I do not want to add this path at all to this commit I am
creating, so please ignore it in the meantime."
> @@ -1088,6 +1090,8 @@ static int parse_and_validate_options(int argc, const char *argv[],
> cleanup_mode = CLEANUP_ALL;
> else
> die(_("Invalid cleanup mode %s"), cleanup_arg);
> + if (skip_intent_to_add)
> + cache_tree_flags = WRITE_TREE_INTENT_TO_ADD_OK;
The name WRITE_TREE_INTENT_TO_ADD_OK says "it is OK to call write-tree
with i-t-a entries in the index, please do not barf", but I think "when
writing a tree, ignore i-t-a entries" would be a more appropriate way to
say the same thing, i.e. WRITE_TREE_IGNORE_INTENT_TO_ADD.
Other than that, I do not see an issue in the implementation of the
patch. It is a separate design level issue if we want to worsen
proliferation of the options, though.
Thanks.
^ permalink raw reply
* Re: [PATCH v3 09/10] clone: allow --branch to take a tag
From: Junio C Hamano @ 2012-01-11 23:57 UTC (permalink / raw)
To: Nguyễn Thái Ngọc Duy; +Cc: git, Jeff King
In-Reply-To: <1326189427-20800-10-git-send-email-pclouds@gmail.com>
Nguyễn Thái Ngọc Duy <pclouds@gmail.com> writes:
> @@ -766,6 +771,14 @@ int cmd_clone(int argc, const char **argv, const char *prefix)
> find_ref_by_name(mapped_refs, head.buf);
> strbuf_release(&head);
>
> + if (!our_head_points_at) {
> + strbuf_addstr(&head, "refs/tags/");
> + strbuf_addstr(&head, option_branch);
> + our_head_points_at =
> + find_ref_by_name(mapped_refs, head.buf);
> + strbuf_release(&head);
> + }
> +
Ok. I think this makes sense.
^ permalink raw reply
* Re: leaky cherry-pick
From: Junio C Hamano @ 2012-01-12 0:00 UTC (permalink / raw)
To: Jeff King
Cc: Ramkumar Ramachandra, Pete Wyckoff, git,
Nguyễn Thái Ngọc
In-Reply-To: <20120111195605.GB12333@sigill.intra.peff.net>
Jeff King <peff@peff.net> writes:
> Maybe this?
>
> diff --git a/attr.c b/attr.c
> index 76b079f..1656db4 100644
> --- a/attr.c
> +++ b/attr.c
> @@ -301,6 +301,7 @@ static void free_attr_elem(struct attr_stack *e)
> }
> free(a);
> }
> + free(e->attrs);
> free(e);
> }
Yeah, that is definitely a leak.
^ permalink raw reply
* Re: [PATCH 2/2] diff --word-diff: use non-whitespace regex by default
From: Tay Ray Chuan @ 2012-01-12 0:52 UTC (permalink / raw)
To: Thomas Rast; +Cc: Git Mailing List, Junio C Hamano
In-Reply-To: <87lipexawp.fsf@thomas.inf.ethz.ch>
Hi,
Thomas, first off, thanks for looking through this.
On Thu, Jan 12, 2012 at 4:05 AM, Thomas Rast <trast@student.ethz.ch> wrote:
> Tay Ray Chuan <rctay89@gmail.com> writes:
>
>> Factor out the comprehensive non-whitespace regex in use by PATTERNS and
>> IPATTERN and use it as the word-diff regex for the default diff driver.
>
> Why?
>
> I seem to recall that the motivation for keeping the original code as-is
> instead of just emulating its behavior with a default regex was that it
> is faster. So disabling the default mode should at least have an
> advantage?
>
> </devils-advocate>
If you're talking about speed, yeah, that's probably true.
But I think it's worthwhile to trade-off performance for a sensible
default. Something like
matrix[a,b,c]
matrix[d,b,c]
gives
matrix[[-a-]{+d+},b,c]
and when we have
ImagineALanguageLikeFoo
ImagineALanguageLikeBar
we get
ImagineALanguageLike[-Foo-]{+Bar+}
(But I cheated. Foo and Bar have no common characters in common; if
they did, the word diff would be messy.)
Both of which seem sensible. From a usability/effectiveness
standpoint, I think it's more useful than what the current word-diff
defaults to - the whole line is taken as a "word", with the pre-image
shown as deleted and the post-image as added; we don't even try to run
LCS on it.
Examples are lifted from:
[1] http://article.gmane.org/gmane.comp.version-control.git/105896
[2] http://article.gmane.org/gmane.comp.version-control.git/105237
--
Cheers,
Ray Chuan
^ permalink raw reply
* Re: [PATCH 1/2] cache-tree: update API to take abitrary flags
From: Nguyen Thai Ngoc Duy @ 2012-01-12 1:20 UTC (permalink / raw)
To: Junio C Hamano; +Cc: git
In-Reply-To: <7vmx9t6bs5.fsf@alter.siamese.dyndns.org>
2012/1/12 Junio C Hamano <gitster@pobox.com>:
> Thanks; this one looks very sensible regardless of what follows (or does
> not follow). Forgot to sign-off?
Deliberately to stop you from using it because I did not test it
carefully. It was created as material for the discussion only. Will
resubmit later.
--
Duy
^ permalink raw reply
* Re: [PATCH v3 06/10] clone: delay cloning until after remote HEAD checking
From: Nguyen Thai Ngoc Duy @ 2012-01-12 1:27 UTC (permalink / raw)
To: Junio C Hamano; +Cc: git, Jeff King
In-Reply-To: <7v8vle6ng3.fsf@alter.siamese.dyndns.org>
2012/1/12 Junio C Hamano <gitster@pobox.com>:
>> @@ -938,6 +940,7 @@ struct transport *transport_get(struct remote *remote, const char *url)
>> char *handler = xmalloc(len + 1);
>> handler[len] = 0;
>> strncpy(handler, url, len);
>> + remote->foreign_vcs = helper;
>> transport_helper_init(ret, handler);
>> }
>
> This I am not sure. What value does "helper" variable have at this point
> in the flow? Wouldn't it be a NULL? Or did you mean "handler"?
Ah yes "handler", my bad.
--
Duy
^ permalink raw reply
* Re: [PATCH 2/3] am: learn passing -b to mailinfo
From: Junio C Hamano @ 2012-01-12 1:35 UTC (permalink / raw)
To: Thomas Rast; +Cc: git
In-Reply-To: <19539098c07a207f3bd24f5a145ba3b6c5e46766.1326312730.git.trast@student.ethz.ch>
Thomas Rast <trast@student.ethz.ch> writes:
> @@ -571,8 +574,8 @@ then
> else
> utf8=-n
> fi
> -if test "$(cat "$dotest/keep")" = t
> -then
> +keep=$(cat "$dotest/keep")
> +if test "$keep" = t
> keep=-k
> fi
Curious.
Who writes 't' to $dotest/keep after this patch is applied?
I also do not want to worry about "echo" portability issues that may come
from an existing
echo "$keep" >"$dotest/keep"
that this patch does not touch.
I suspect that this patch was not tested in a way to exercise this
codepath; shell would have barfed when seeing the lack of "then" here, no?
^ permalink raw reply
* Re: [PATCH 1/2] t9200: On MSYS, do not pass Windows-style paths to CVS
From: Junio C Hamano @ 2012-01-12 2:06 UTC (permalink / raw)
To: Sebastian Schuberth; +Cc: msysgit, git
In-Reply-To: <4F0D544E.6010105@gmail.com>
Sebastian Schuberth <sschuberth@gmail.com> writes:
> For details, see the commit message of 4114156ae9. Note that while using
> $PWD as part of GIT_DIR is not required here, it does no harm and it is
> more consistent. In addition, on MSYS using an environment variable should
> be slightly faster than spawning an external executable.
Thanks. It seems that the "Dos and Dont's" from t/README needs to be
stressed a bit more strongly.
^ permalink raw reply
* Re: git diff <file> HEAD^:<file> error message
From: Junio C Hamano @ 2012-01-12 2:26 UTC (permalink / raw)
To: Carlos Martín Nieto; +Cc: git
In-Reply-To: <20120111111831.GB15232@beez.lab.cmartin.tk>
Carlos Martín Nieto <cmn@elego.de> writes:
> I was trying to figure out why running
>
> git diff HEAD^:RelNotes RelNotes
>
> gives the expected output (on maint it tells me that the stable
> version changed from 1.7.8.3 to 1.7.8.4) but swapping the arguments
> doesn't.
>
> git diff RelNotes HEAD^:RelNotes
>
> doesn't show the opposite patch ...
That comes from the general argument parsing rules of Git, namely, global
options (e.g. --paginate) first, then subcommand name, followed by dashed
options, revs and finally the paths. Once you give "RelNotes", which
cannot be a rev, you cannot give a rev.
We _could_ special case the rule for "diff", but we simply didn't bother,
as the resulting code (and the implications of special casing) would be
too ugly to live to support such a corner case usage, especially when you
could always say "-R" to reverse the output.
^ permalink raw reply
* Re: [PATCH 1/2] get_sha1_with_context: report features used in resolution
From: Junio C Hamano @ 2012-01-12 2:36 UTC (permalink / raw)
To: Jeff King; +Cc: Carlos Martín Nieto, git, Albert Astals Cid
In-Reply-To: <20120111194210.GA12441@sigill.intra.peff.net>
Jeff King <peff@peff.net> writes:
> Most callers generally treat get_sha1 as a black box, giving
> it a string from the user and expecting to get a sha1 in
> return. The get_sha1_with_context function gives callers
> more information about what happened while resolving the
> object name so they can make better decisions about how to
> use the result. We currently use this only to provide
> information about the path entry used to find a blob.
>
> We don't currently provide any information about the
> resolution rules that were used to reach the final object.
> Some callers may want these in order to enforce a policy
> that a particular subset of the lookup rules are used (e.g.,
> when serving remote requests).
>
> This patch adds a set of bit-fields that document the use of
> particular features during an object lookup.
>
> Signed-off-by: Jeff King <peff@peff.net>
> ---
> The diffstat looks a little scary, but it is mostly just the internal
> get_sha1 functions learning to pass the object_context around.
Hmm, shouldn't this also cover peel_to_type()? That would have made it
also apply to the maintenance track.
^ permalink raw reply
* Re: [PATCH] archive: re-allow HEAD:Documentation on a remote invocation
From: Junio C Hamano @ 2012-01-12 2:46 UTC (permalink / raw)
To: Carlos Martín Nieto; +Cc: Jeff King, git, Albert Astals Cid
In-Reply-To: <1326283958-30271-1-git-send-email-cmn@elego.de>
Carlos Martín Nieto <cmn@elego.de> writes:
> The tightening done in (ee27ca4a: archive: don't let remote clients
> get unreachable commits, 2011-11-17) went too far and disallowed
> HEAD:Documentation as it would try to find "HEAD:Documentation" as a
> ref.
I do not think it went too far. Actually we discussed this exact issue
when the topic was cooking, and saw no objections. The commit in question
itself advertises this restriction.
Why are we loosening it now? I do not see a compelling reason to do so.
> Only DWIM the "HEAD" part to see if it exists as a ref. Once we're
> sure that we've been given a valid ref, we follow the normal code
> path. This still disallows attempts to access commits which are not
> branch tips.
>
> Signed-off-by: Carlos Martín Nieto <cmn@elego.de>
> ---
> AFAICT this should still be safe. Using HEAD^:Documentation or
> <sha1>:Documentation still complains that HEAD^ and <sha1> aren't
> refs.
Having said that, I think I agree this is a safe thing to do, _if_ we want
to loosen it.
^ permalink raw reply
* Re: [PATCH 1/2] get_sha1_with_context: report features used in resolution
From: Jeff King @ 2012-01-12 2:51 UTC (permalink / raw)
To: Junio C Hamano; +Cc: Carlos Martín Nieto, git, Albert Astals Cid
In-Reply-To: <7vmx9t4pgj.fsf@alter.siamese.dyndns.org>
On Wed, Jan 11, 2012 at 06:36:12PM -0800, Junio C Hamano wrote:
> Jeff King <peff@peff.net> writes:
>
> > Most callers generally treat get_sha1 as a black box, giving
> > it a string from the user and expecting to get a sha1 in
> > return. The get_sha1_with_context function gives callers
> > more information about what happened while resolving the
> > object name so they can make better decisions about how to
> > use the result. We currently use this only to provide
> > information about the path entry used to find a blob.
> >
> > We don't currently provide any information about the
> > resolution rules that were used to reach the final object.
> > Some callers may want these in order to enforce a policy
> > that a particular subset of the lookup rules are used (e.g.,
> > when serving remote requests).
> >
> > This patch adds a set of bit-fields that document the use of
> > particular features during an object lookup.
> >
> > Signed-off-by: Jeff King <peff@peff.net>
> > ---
> > The diffstat looks a little scary, but it is mostly just the internal
> > get_sha1 functions learning to pass the object_context around.
>
> Hmm, shouldn't this also cover peel_to_type()? That would have made it
> also apply to the maintenance track.
I don't see how peel_to_type is relevant. As far as get_sha1 is
concerned, the interesting thing is actually calling peel_onion. It does
get the context passed to it in my patch, but I didn't bother marking
that the peel feature was used (because it wasn't relevant to the policy
I wanted to implement in the follow-on patch).
But we could pretty easily mark the use of the peel feature, too.
I'm not sure what you mean about the maintenance track, though. AFAICT,
we don't separately call peel_to_type, but just potentially use it as
part of get_sha1_with_context. Am I missing something?
-Peff
^ permalink raw reply
* Re: [PATCH RFC] commit: allow to commit even if there are intent-to-add entries
From: Nguyen Thai Ngoc Duy @ 2012-01-12 2:53 UTC (permalink / raw)
To: Junio C Hamano; +Cc: Jonathan Nieder, git, Jeff King, Will Palmer
In-Reply-To: <7vr4z654m3.fsf@alter.siamese.dyndns.org>
2012/1/12 Junio C Hamano <gitster@pobox.com>:
> Jonathan Nieder <jrnieder@gmail.com> writes:
>
>>> When running "commit" and "status" with files marked with "intent to add",
>>> I think there are three possible interpretations of what the user
>>> wants to do.
>> [ (1) thanks for stopping me, I had forgotten about that file;
>> (2) I changed my mind, please leave that file out; or (3) please
>> dwim and add the file ]
>>
>> I think (3) was a trick --- no one that does not use the "-a" option
>> would want that. :)
>
> I really wish it were the case, but I doubt it.
>
> People from other VCS background seem to still think that "commit" without
> paths should commit everything; after getting told that "what you add to
> the index is what you will commit", I can easily see this commint: but but
> but I told Git that I _want_ to add with -N! Why aren't you committing it?
I see "-N" just as an indication, not really an "add" operation.
Perhaps update-index is a better place for it.
>> (2) makes intent-to-add entries just like any other tracked index
>> entry with some un-added content.
>
> You are comparing files edited in the working tree without the user
> telling anything about them to Git (both tracked and untracked) and files
> the user explicitly told Git that the user hasn't made up her mind
> about. Why is it a good thing to make the latter behave "just like any
> other"?
The way I see this flag is "include these files in my diff in addition
to tracked files", and therefore should not have any effects at commit
time. I might turn some of those extra files to tracked some time if I
want to commit.
--
Duy
^ permalink raw reply
* Re: [PATCH] archive: re-allow HEAD:Documentation on a remote invocation
From: Jeff King @ 2012-01-12 2:54 UTC (permalink / raw)
To: Junio C Hamano; +Cc: Carlos Martín Nieto, git, Albert Astals Cid
In-Reply-To: <7vipkh4oyn.fsf@alter.siamese.dyndns.org>
On Wed, Jan 11, 2012 at 06:46:56PM -0800, Junio C Hamano wrote:
> Carlos Martín Nieto <cmn@elego.de> writes:
>
> > The tightening done in (ee27ca4a: archive: don't let remote clients
> > get unreachable commits, 2011-11-17) went too far and disallowed
> > HEAD:Documentation as it would try to find "HEAD:Documentation" as a
> > ref.
>
> I do not think it went too far. Actually we discussed this exact issue
> when the topic was cooking, and saw no objections. The commit in question
> itself advertises this restriction.
I think you and I discussed it off list (I originally took this off-list
because the original issue did have some security implications). So I
don't think people necessarily had a chance to object.
> Why are we loosening it now? I do not see a compelling reason to do so.
I see it the opposite way. People are clearly using the "$ref:$path"
syntax. So why would we restrict them from doing so? There are no
security implications (i.e., they could always just grab $ref and
extract $path themselves). In my view, ee27ca4a was over-eager in its
restrictions because I wanted it to be simple and close the hole. Now we
can take our time adding more code to loosen it.
-Peff
^ permalink raw reply
* Re: [PATCH] archive: re-allow HEAD:Documentation on a remote invocation
From: Jeff King @ 2012-01-12 2:59 UTC (permalink / raw)
To: Junio C Hamano; +Cc: Carlos Martín Nieto, git, Albert Astals Cid
In-Reply-To: <20120112025445.GB25365@sigill.intra.peff.net>
On Wed, Jan 11, 2012 at 09:54:45PM -0500, Jeff King wrote:
> On Wed, Jan 11, 2012 at 06:46:56PM -0800, Junio C Hamano wrote:
>
> > Carlos Martín Nieto <cmn@elego.de> writes:
> >
> > > The tightening done in (ee27ca4a: archive: don't let remote clients
> > > get unreachable commits, 2011-11-17) went too far and disallowed
> > > HEAD:Documentation as it would try to find "HEAD:Documentation" as a
> > > ref.
> >
> > I do not think it went too far. Actually we discussed this exact issue
> > when the topic was cooking, and saw no objections. The commit in question
> > itself advertises this restriction.
>
> I think you and I discussed it off list (I originally took this off-list
> because the original issue did have some security implications). So I
> don't think people necessarily had a chance to object.
Here is the only on-list discussion:
http://article.gmane.org/gmane.comp.version-control.git/186366
Quoted below:
>> * jk/maint-1.6.2-upload-archive (2011-11-21) 1 commit
>> - archive: don't let remote clients get unreachable commits
>> (this branch is used by jk/maint-upload-archive.)
>>
>> * jk/maint-upload-archive (2011-11-21) 1 commit
>> - Merge branch 'jk/maint-1.6.2-upload-archive' into
>> jk/maint-upload-archive
>> (this branch uses jk/maint-1.6.2-upload-archive.)
>>
>> Will merge to 'next' after taking another look.
>
> Thanks. I also have some followup patches to re-loosen to at least
> trees reachable from refs. Do you want to leave the tightening to
> the maint track, and then consider the re-loosening for master?
I was planning to first have the really tight version graduate to
'master' and ship it in 1.7.9, while possibly merging that to 1.7.8.X
series. If we hear complaints from real users in the meantime before
or after such releases, we could apply loosening patch on top of these
topics and call them "regression fix", but I have been assuming that
nobody would have been using this backdoor for anything that really
matters.
So now we have heard a complaint. :)
-Peff
^ permalink raw reply
* Re: [PATCH v3 07/10] clone: --branch=<branch> always means refs/heads/<branch>
From: Nguyen Thai Ngoc Duy @ 2012-01-12 3:00 UTC (permalink / raw)
To: Junio C Hamano; +Cc: git, Jeff King
In-Reply-To: <7v4nw26mbe.fsf@alter.siamese.dyndns.org>
2012/1/12 Junio C Hamano <gitster@pobox.com>:
> Nguyễn Thái Ngọc Duy <pclouds@gmail.com> writes:
>
>> It does not make sense to look outside refs/heads for HEAD's target
>> (src_ref_prefix can be set to "refs/" if --mirror is used) because ref
>> code only allows symref in form refs/heads/...
>>
>> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
>> ---
>> builtin/clone.c | 4 ++--
>> 1 files changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/builtin/clone.c b/builtin/clone.c
>> index 05224d7..960242f 100644
>> --- a/builtin/clone.c
>> +++ b/builtin/clone.c
>> @@ -463,7 +463,7 @@ static void update_remote_refs(const struct ref *refs,
>> static void update_head(const struct ref *our, const struct ref *remote,
>> const char *msg)
>> {
>> - if (our) {
>> + if (our && !prefixcmp(our->name, "refs/heads/")) {
>> /* Local default branch link */
>> create_symref("HEAD", our->name, NULL);
>> if (!option_bare) {
>
> I think this makes sense. In the following three casees:
>
> - When cloning without --branch, if we could not find a branch that match
> HEAD at the remote;
>
> - When cloning with --branch, if we did not see such a branch at the
> remote; or
>
> - When cloning from an empty repository
>
> we come here with "our" set to NULL. Additionally, if the remote HEAD
> points outside refs/heads/ and the transport could detect that case
> (e.g. a helper that reads from ls-remote output), we can see our->name
> outside refs/heads/. Is there any other case where our is not NULL and
> our->name does not start with refs/heads/?
No, this patch pretty much guarantees that our->name must be inside
refs/heads, unless remote's HEAD points outside, which makes it a
broken remote and git-upload-pack should refuse to serve in the first
place. Until --branch=tag comes into the picture.
> The "else if (remote)" clause (not shown, outside the context) that
> follows still has comments that says it is a case for "Source had detached
> HEAD pointing somewhere", and needs to be adjusted for this change, no? It
> is "(1) we know the HEAD points at a non-branch, (2) HEAD may point at a
> branch but we do not know which one, or (3) we asked for a specific branch
> but it did not exist there" (cloning from an empty will have NULL in
> remote and the comment would not apply to that case).
Yes.
--
Duy
^ permalink raw reply
* Re: [PATCH] archive: re-allow HEAD:Documentation on a remote invocation
From: Junio C Hamano @ 2012-01-12 3:03 UTC (permalink / raw)
To: Jeff King; +Cc: Carlos Martín Nieto, git, Albert Astals Cid
In-Reply-To: <20120112025445.GB25365@sigill.intra.peff.net>
Jeff King <peff@peff.net> writes:
> I see it the opposite way. People are clearly using the "$ref:$path"
> syntax. So why would we restrict them from doing so? There are no
> security implications (i.e., they could always just grab $ref and
> extract $path themselves). In my view, ee27ca4a was over-eager in its
> restrictions because I wanted it to be simple and close the hole. Now we
> can take our time adding more code to loosen it.
Ok, so it is more like a partial revert of whatever we did. In that case,
I'd take CMN's patch to limit the extent of the changes, as it more
closely matches the spirit of the original ee27ca4 (archive: don't let
remote clients get unreachable commits, 2011-11-17) that singled out and
catered to the need of "archive" command alone. It is already part of the
v1.7.8.1 release, so I would prefer a change to be stupid and simple.
^ permalink raw reply
* Re: leaky cherry-pick
From: Jeff King @ 2012-01-12 3:05 UTC (permalink / raw)
To: Junio C Hamano
Cc: Ramkumar Ramachandra, Pete Wyckoff, git,
Nguyễn Thái Ngọc
In-Reply-To: <7v7h0x6b89.fsf@alter.siamese.dyndns.org>
On Wed, Jan 11, 2012 at 04:00:38PM -0800, Junio C Hamano wrote:
> Yeah, that is definitely a leak.
Here it is with a commit message. "valgrind git cherry-pick" reports no
leaks in the attr code after this (but unfortunately still lots of leaks
from unpack-trees).
-Peff
-- >8 --
Subject: [PATCH] attr: fix leak in free_attr_elem
This function frees the individual "struct match_attr"s we
have allocated, but forgot to free the array holding their
pointers, leading to a very minor memory leak.
Signed-off-by: Jeff King <peff@peff.net>
---
attr.c | 1 +
1 files changed, 1 insertions(+), 0 deletions(-)
diff --git a/attr.c b/attr.c
index 96eda0e..303751f 100644
--- a/attr.c
+++ b/attr.c
@@ -301,6 +301,7 @@ static void free_attr_elem(struct attr_stack *e)
}
free(a);
}
+ free(e->attrs);
free(e);
}
--
1.7.9.rc0.33.gd3c17
^ permalink raw reply related
* Re: [PATCH RFC] commit: allow to commit even if there are intent-to-add entries
From: Junio C Hamano @ 2012-01-12 3:05 UTC (permalink / raw)
To: Nguyen Thai Ngoc Duy; +Cc: Jonathan Nieder, git, Jeff King, Will Palmer
In-Reply-To: <CACsJy8CHOuZMP4hWp6Lb_TbdfwSgofSg_0tDZ4oDcD0veie2Dw@mail.gmail.com>
Nguyen Thai Ngoc Duy <pclouds@gmail.com> writes:
>> I really wish it were the case, but I doubt it.
>>
>> People from other VCS background seem to still think that "commit" without
>> paths should commit everything; after getting told that "what you add to
>> the index is what you will commit", I can easily see this commint: but but
>> but I told Git that I _want_ to add with -N! Why aren't you committing it?
>
> I see "-N" just as an indication, not really an "add" operation.
But the thing is, what _you_, who knows a bit more than these new people
about Git, see does not matter, because they would not listen to you.
^ permalink raw reply
* Re: [PATCHv3 10/13] credentials: add "cache" helper
From: Jeff King @ 2012-01-12 3:07 UTC (permalink / raw)
To: Jonathan Nieder; +Cc: Junio C Hamano, git
In-Reply-To: <20120111235009.GB30243@burratino>
On Wed, Jan 11, 2012 at 05:50:10PM -0600, Jonathan Nieder wrote:
> Thanks again for the fix. Here's another quick nit.
My favorite form for nits to come in: a patch.
> -- >8 --
> Subject: unix-socket: do not let close() or chdir() clobber errno during cleanup
> [...]
Looks good to me. Thanks.
-Peff
^ permalink raw reply
* Re: [PATCH] archive: re-allow HEAD:Documentation on a remote invocation
From: Jeff King @ 2012-01-12 3:10 UTC (permalink / raw)
To: Junio C Hamano; +Cc: Carlos Martín Nieto, git, Albert Astals Cid
In-Reply-To: <7vehv54o6v.fsf@alter.siamese.dyndns.org>
On Wed, Jan 11, 2012 at 07:03:36PM -0800, Junio C Hamano wrote:
> Jeff King <peff@peff.net> writes:
>
> > I see it the opposite way. People are clearly using the "$ref:$path"
> > syntax. So why would we restrict them from doing so? There are no
> > security implications (i.e., they could always just grab $ref and
> > extract $path themselves). In my view, ee27ca4a was over-eager in its
> > restrictions because I wanted it to be simple and close the hole. Now we
> > can take our time adding more code to loosen it.
>
> Ok, so it is more like a partial revert of whatever we did. In that case,
> I'd take CMN's patch to limit the extent of the changes, as it more
> closely matches the spirit of the original ee27ca4 (archive: don't let
> remote clients get unreachable commits, 2011-11-17) that singled out and
> catered to the need of "archive" command alone. It is already part of the
> v1.7.8.1 release, so I would prefer a change to be stupid and simple.
For a maint release, I am OK with that. In the long term, I'd rather my
patches go onto master (either for 1.7.9 or for later), as I think they
are the right way to do it.
-Peff
^ permalink raw reply
* Re: [PATCH] archive: re-allow HEAD:Documentation on a remote invocation
From: Junio C Hamano @ 2012-01-12 3:20 UTC (permalink / raw)
To: Jeff King; +Cc: Carlos Martín Nieto, git, Albert Astals Cid
In-Reply-To: <20120112031022.GC26363@sigill.intra.peff.net>
Jeff King <peff@peff.net> writes:
> For a maint release, I am OK with that. In the long term, I'd rather my
> patches go onto master (either for 1.7.9 or for later), as I think they
> are the right way to do it.
I would prefer not to use it in 1.7.9, as CMN's patch will make its only
immediate user disappear. I think we are in agreement that the change to
carry context around is a longer term thing, and when we apply it, the
safety we added to "archive" will become simpler.
^ permalink raw reply
* [PATCH] tree_entry_interesting: make recursive mode default
From: Nguyễn Thái Ngọc Duy @ 2012-01-12 4:09 UTC (permalink / raw)
To: git
Cc: Junio C Hamano, Jonathan Nieder, Linus Torvalds,
Nguyễn Thái Ngọc Duy
In-Reply-To: <20120111063104.GA3153@burratino>
There is a bit of history behind this. Some time ago, t_e_i() only
supported prefix matching. diff-tree supported recursive and
non-recursive mode but it did not make any difference to prefix
matching.
Later on, t_e_i() gained limited recursion support to unify a similar
matching code used by git-grep. It introduced two new fields in struct
pathspec: max_depth and recursive. "recursive" field functions as a
feature switch so that this feature is off by default.
Some time after that, t_e_i() further gained wildcard support. With
wildcard matching, recursive and non-recursive diff-tree
mattered. "recursive" field was reused to distinguish recursion in
diff-tree.
This choice has a side effect that by default wildcard matching is in
non-recursive mode, which is not true. All current call sites except
"diff-tree without -r" (grep, traverse_commit_list, tree-walk and
general tree diff) prefer recursive mode.
This patch decouples the use of recursive field. The max_depth feature
switch is now controlled by max_depth_valid field. diff-tree recursion
is controlled by nonrecursive_diff_tree, which makes it recursive by
default.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
Junio, log/diff family takes wildcards just fine (even before this
patch, just less reliable). If it's not mentioned in release notes
before, perhaps it's worth a line in 1.7.9 annoucement.
builtin/grep.c | 2 +-
cache.h | 3 ++-
dir.c | 4 ++--
t/t4010-diff-pathspec.sh | 8 ++++++++
tree-diff.c | 3 +--
tree-walk.c | 8 ++++----
6 files changed, 18 insertions(+), 10 deletions(-)
diff --git a/builtin/grep.c b/builtin/grep.c
index 9ce064a..c081bf4 100644
--- a/builtin/grep.c
+++ b/builtin/grep.c
@@ -1051,7 +1051,7 @@ int cmd_grep(int argc, const char **argv, const char *prefix)
paths = get_pathspec(prefix, argv + i);
init_pathspec(&pathspec, paths);
pathspec.max_depth = opt.max_depth;
- pathspec.recursive = 1;
+ pathspec.max_depth_valid = 1;
if (show_in_pager && (cached || list.nr))
die(_("--open-files-in-pager only works on the worktree"));
diff --git a/cache.h b/cache.h
index 10afd71..f58e05a 100644
--- a/cache.h
+++ b/cache.h
@@ -526,7 +526,8 @@ struct pathspec {
const char **raw; /* get_pathspec() result, not freed by free_pathspec() */
int nr;
unsigned int has_wildcard:1;
- unsigned int recursive:1;
+ unsigned int max_depth_valid:1;
+ unsigned int nonrecursive_diff_tree:1;
int max_depth;
struct pathspec_item {
const char *match;
diff --git a/dir.c b/dir.c
index 0a78d00..5af3567 100644
--- a/dir.c
+++ b/dir.c
@@ -258,7 +258,7 @@ int match_pathspec_depth(const struct pathspec *ps,
int i, retval = 0;
if (!ps->nr) {
- if (!ps->recursive || ps->max_depth == -1)
+ if (!ps->max_depth_valid || ps->max_depth == -1)
return MATCHED_RECURSIVELY;
if (within_depth(name, namelen, 0, ps->max_depth))
@@ -275,7 +275,7 @@ int match_pathspec_depth(const struct pathspec *ps,
if (seen && seen[i] == MATCHED_EXACTLY)
continue;
how = match_pathspec_item(ps->items+i, prefix, name, namelen);
- if (ps->recursive && ps->max_depth != -1 &&
+ if (ps->max_depth_valid && ps->max_depth != -1 &&
how && how != MATCHED_FNMATCH) {
int len = ps->items[i].len;
if (name[len] == '/')
diff --git a/t/t4010-diff-pathspec.sh b/t/t4010-diff-pathspec.sh
index fbc8cd8..af5134b 100755
--- a/t/t4010-diff-pathspec.sh
+++ b/t/t4010-diff-pathspec.sh
@@ -48,6 +48,14 @@ test_expect_success \
compare_diff_raw current expected'
cat >expected <<\EOF
+:100644 100644 766498d93a4b06057a8e49d23f4068f1170ff38f 0a41e115ab61be0328a19b29f18cdcb49338d516 M path1/file1
+EOF
+test_expect_success \
+ '"*file1" should show path1/file1' \
+ 'git diff-index --cached $tree -- "*file1" >current &&
+ compare_diff_raw current expected'
+
+cat >expected <<\EOF
:100644 100644 766498d93a4b06057a8e49d23f4068f1170ff38f 0a41e115ab61be0328a19b29f18cdcb49338d516 M file0
EOF
test_expect_success \
diff --git a/tree-diff.c b/tree-diff.c
index 28ad6db..164693f 100644
--- a/tree-diff.c
+++ b/tree-diff.c
@@ -137,8 +137,7 @@ int diff_tree(struct tree_desc *t1, struct tree_desc *t2,
enum interesting t2_match = entry_not_interesting;
/* Enable recursion indefinitely */
- opt->pathspec.recursive = DIFF_OPT_TST(opt, RECURSIVE);
- opt->pathspec.max_depth = -1;
+ opt->pathspec.nonrecursive_diff_tree = !DIFF_OPT_TST(opt, RECURSIVE);
strbuf_init(&base, PATH_MAX);
strbuf_add(&base, base_str, baselen);
diff --git a/tree-walk.c b/tree-walk.c
index f82dba6..04f1b81 100644
--- a/tree-walk.c
+++ b/tree-walk.c
@@ -588,7 +588,7 @@ enum interesting tree_entry_interesting(const struct name_entry *entry,
entry_not_interesting : all_entries_not_interesting;
if (!ps->nr) {
- if (!ps->recursive || ps->max_depth == -1)
+ if (!ps->max_depth_valid || ps->max_depth == -1)
return all_entries_interesting;
return within_depth(base->buf + base_offset, baselen,
!!S_ISDIR(entry->mode),
@@ -609,7 +609,7 @@ enum interesting tree_entry_interesting(const struct name_entry *entry,
if (!match_dir_prefix(base_str, match, matchlen))
goto match_wildcards;
- if (!ps->recursive || ps->max_depth == -1)
+ if (!ps->max_depth_valid || ps->max_depth == -1)
return all_entries_interesting;
return within_depth(base_str + matchlen + 1,
@@ -634,7 +634,7 @@ enum interesting tree_entry_interesting(const struct name_entry *entry,
* Match all directories. We'll try to
* match files later on.
*/
- if (ps->recursive && S_ISDIR(entry->mode))
+ if (!ps->nonrecursive_diff_tree && S_ISDIR(entry->mode))
return entry_interesting;
}
@@ -662,7 +662,7 @@ match_wildcards:
* Match all directories. We'll try to match files
* later on.
*/
- if (ps->recursive && S_ISDIR(entry->mode))
+ if (!ps->nonrecursive_diff_tree && S_ISDIR(entry->mode))
return entry_interesting;
}
return never_interesting; /* No matches */
--
1.7.3.1.256.g2539c.dirty
^ permalink raw reply related
* Re: [PATCH] tree_entry_interesting: make recursive mode default
From: Junio C Hamano @ 2012-01-12 5:04 UTC (permalink / raw)
To: Nguyễn Thái Ngọc Duy
Cc: git, Jonathan Nieder, Linus Torvalds
In-Reply-To: <1326341371-16628-1-git-send-email-pclouds@gmail.com>
Nguyễn Thái Ngọc Duy <pclouds@gmail.com> writes:
> There is a bit of history behind this. Some time ago, t_e_i() only
> supported prefix matching. diff-tree supported recursive and
> non-recursive mode but it did not make any difference to prefix
> matching.
>
> Later on, t_e_i() gained limited recursion support to unify a similar
> matching code used by git-grep. It introduced two new fields in struct
> pathspec: max_depth and recursive. "recursive" field functions as a
> feature switch so that this feature is off by default.
>
> Some time after that, t_e_i() further gained wildcard support. With
> wildcard matching, recursive and non-recursive diff-tree
> mattered. "recursive" field was reused to distinguish recursion in
> diff-tree.
>
> This choice has a side effect that by default wildcard matching is in
> non-recursive mode, which is not true. All current call sites except
> "diff-tree without -r" (grep, traverse_commit_list, tree-walk and
> general tree diff) prefer recursive mode.
>
> This patch decouples the use of recursive field. The max_depth feature
> switch is now controlled by max_depth_valid field. diff-tree recursion
> is controlled by nonrecursive_diff_tree, which makes it recursive by
> default.
Thanks, but I am curious about two (and a half) things.
- The "max_depth" option has perfectly good and natural "invalid"
sentinel values (i.e. 0 or negative). Why do we need a separate
bitfield?
- Special casing the non-recursive mode of diff-tree is perfectly
acceptable, but nonrecursive_diff_tree does not sound like a very good
name for it for two reasons. Perhaps there may be other users that want
the "surface only" behaviour, so having "diff_tree" in the name limits
its future (re)use. Also an option that is named negatively inevitably
invites "if (!opt->non_whatever)" double negative. Can we come up with
a better name, perhaps "onelevel_only" or something?
- Shouldn't "onelevel_only" be the same as limiting to a single depth
with "max_depth"?
^ permalink raw reply
* Re: [PATCH] tree_entry_interesting: make recursive mode default
From: Nguyen Thai Ngoc Duy @ 2012-01-12 5:44 UTC (permalink / raw)
To: Junio C Hamano; +Cc: git, Jonathan Nieder, Linus Torvalds
In-Reply-To: <7v1ur54ikk.fsf@alter.siamese.dyndns.org>
2012/1/12 Junio C Hamano <gitster@pobox.com>:
> Thanks, but I am curious about two (and a half) things.
>
> - The "max_depth" option has perfectly good and natural "invalid"
> sentinel values (i.e. 0 or negative). Why do we need a separate
> bitfield?
We want infinite recursion by default, by max_depth default value is
0, which is non-recursive.
> - Special casing the non-recursive mode of diff-tree is perfectly
> acceptable, but nonrecursive_diff_tree does not sound like a very good
> name for it for two reasons. Perhaps there may be other users that want
> the "surface only" behaviour, so having "diff_tree" in the name limits
> its future (re)use. Also an option that is named negatively inevitably
> invites "if (!opt->non_whatever)" double negative. Can we come up with
> a better name, perhaps "onelevel_only" or something?
I'm bad at naming, any other names are welcome. onelevel_only sounds good.
> - Shouldn't "onelevel_only" be the same as limiting to a single depth
> with "max_depth"?
I thought of that but not sure it's equivalent to max_depth == 1 or
max_depth == 0, so I separated it for safety and clarity. max_depth
feature is driven by git-grep and there were a few interpretations how
it should behave last time, so I'm not sure if its behavior may change
in future.
--
Duy
^ permalink raw reply
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox