* Re: [PATCH 0/5] ignore SIG{INT,QUIT} when launching editor
From: Paul Fox @ 2012-12-01 15:48 UTC (permalink / raw)
To: Jeff King; +Cc: git, Junio C Hamano, Krzysztof Mazur
In-Reply-To: <20121130223943.GA27120@sigill.intra.peff.net>
jeff wrote:
> This is a re-roll of the pf/editor-ignore-sigint series.
>
> There are two changes from the original:
>
> 1. We ignore both SIGINT and SIGQUIT for "least surprise" compared to
> system(3).
>
> 2. We now use "code + 128" to look for signal death (instead of
> WTERMSIG), as per run-command's documentation on how it munges the
> code.
this series all looks good to me. thanks for re- and re-re-rolling.
paul
>
> People mentioned some buggy editors which go into an infinite EIO loop
> when their parent dies due to SIGQUIT. That should be a non-issue now,
> as we will be ignoring SIGQUIT. And even if you could replicate it
> (e.g., with another signal) those programs should be (and reportedly
> have been) fixed. It is not git's job to babysit its child processes.
>
> The patches are:
>
> [1/5]: run-command: drop silent_exec_failure arg from wait_or_whine
> [2/5]: launch_editor: refactor to use start/finish_command
> [3/5]: launch_editor: ignore terminal signals while editor has control
> [4/5]: run-command: do not warn about child death from terminal
> [5/5]: launch_editor: propagate signals from editor to git
>
> Since this can be thought of as "act more like system(3)", I wondered
> whether the signal-ignore logic should be moved into run-command, or
> even used by default for blocking calls to run_command (which are
> basically our version of system(3)). But it is detrimental in the common
> case that the child is not taking control of the terminal, and is just
> an implementation detail (e.g., we call "git update-ref" behind the
> scenes, but the user does not know or care). If they hit ^C during such
> a run and we are ignoring SIGINT, then either:
>
> 1. we will notice the child died by signal and report an
> error in the subprocess rather than just dying; the end result is
> similar, but the error is unnecessarily confusing
>
> 2. we do not bother to check the child's return code (because we do
> not care whether the child succeeded or not, like a "gc --auto");
> we end up totally ignoring the user's request to abort the
> operation
>
> So I do not think we care about this behavior except for launching the
> editor. And the signal-propagation behavior of 5/5 is really so weirdly
> editor-specific (because it is about behaving well whether the child
> blocks signals or not).
>
> -Peff
=---------------------
paul fox, pgf@foxharp.boston.ma.us (arlington, ma, where it's 24.8 degrees)
^ permalink raw reply
* Re: [RFC] remove/deprecate 'submodule init' and 'sync'
From: Jens Lehmann @ 2012-12-01 15:42 UTC (permalink / raw)
To: W. Trevor King
Cc: Phil Hord, Git, Heiko Voigt, Junio C Hamano, Jeff King,
Shawn Pearce, Nahor
In-Reply-To: <20121201124842.GA32291@odin.tremily.us>
Am 01.12.2012 13:48, schrieb W. Trevor King:
> On Fri, Nov 30, 2012 at 06:52:22PM -0500, Phil Hord wrote:
>> If I never 'submodule init' a submodule, it does not get visited by 'git submodule foreach', among others. I think some people use this behavior explicitly.
>
> This is something I'll fix while working up a trial patch. Currently cmd_update calls module_clone if the <submodule>/.git does not exist. This should probably happen in each command (in a wrapper around module_list?). It's possible that module_list itself would need some work, but I haven't absorbed its implementation yet [1].
Please do not fix it, this is a feature. "update" is the only command
where that should happen (and only if "url" is set in .git/config, as
I explained in my other mail). So everything should be fine here.
^ permalink raw reply
* Re: [RFC] remove/deprecate 'submodule init' and 'sync'
From: Jens Lehmann @ 2012-12-01 15:38 UTC (permalink / raw)
To: W. Trevor King
Cc: Phil Hord, Git, Heiko Voigt, Junio C Hamano, Jeff King,
Shawn Pearce, Nahor
In-Reply-To: <20121130175309.GA718@odin.tremily.us>
Am 30.11.2012 18:53, schrieb W. Trevor King:
> In my v5 patch, I check for submodule.<name>.remote first in the usual
> `git config` files. If I don't find what I'm looking for I fall back
> on .gitmodules (basically Jens' suggestion). However, my initial
> copying-to-.git/config approach was mostly done to mimic existing
> configuration handling in git-submodule.sh. Since I agree with Jens
> on configuration precendence, and I now had two options to read
> (.branch and .remote), I thought I'd pull the logic out into its own
> function (code included at the end). While I was shifting the
> existing submodule config handling over to my new function, I noticed
> that with this logic, `submodule init` doesn't really do anything
> important anymore. Likewise for `submodule sync`, which seems to be
> quite similar to `init`.
You need to handle the 'url' setting differently. While I think the
'update' setting should not be copied into .git/config at all
(because it makes it impossible for upstream to change that later
without the user copying that himself as 'sync' doesn't do that) the
'url' setting in .git/config has two important implications:
1) It tells the submodule commands that the user wants to have that
submodule populated (which is done in a subsequent "update" after
"init" copied the url there).
2) It can be used to follow moving upstreams (think of checking out
an earlier commit before the upstream was moved, you won't be able
to clone it from there without having the new setting persist).
And which repository you follow is a matter of trust, so the extra
"git submodule sync" in that case is a good thing to have.
So I believe 'url' is the only setting that should be copied into
.git/config while all the others shouldn't.
> What to do about this? `init` has been around for a while, so we
> can't just remove it (maybe in 2.0?). Leaving it in place is not
> really a problem though, it just means that the user is locking in the
> current .gitmodules configuration (as Jens pointed out with respect to
> .branch).
We still need those commands to set and update the "url" setting.
> ---
> #
> # Print a submodule configuration setting
> #
> # $1 = submodule name
> # $2 = option name
> # $3 = default value
> #
> # Checks in the usual git-config places first (for overrides),
> # otherwise it falls back on .gitmodules. This allows you to
> # distribute project-wide defaults in .gitmodules, while still
> # customizing individual repositories if necessary. If the option is
> # not in .gitmodules either, print a default value.
> #
> get_submodule_config()
> {
> name="$1"
> option="$2"
> default="$3"
> value=$(git config submodule."$name"."$option")
> if test -z "$value"
> then
> value=$(git config -f .gitmodules submodule."$name"."$option")
> fi
> printf '%s' "${value:-$default}"
> }
Something like that makes sense. You can use it for the settings you add
first and we can then reuse that for 'update' in a separate patch later.
^ permalink raw reply
* Also close config file handle when leaving git_config_set_multivar_in_file() early
From: 乙酸鋰 @ 2012-12-01 14:01 UTC (permalink / raw)
To: git
[-- Attachment #1: Type: text/plain, Size: 221 bytes --]
Hi,
This patch fixes the captioned problem.
It is needed because our program statically link git.
In this case, do not assume a fail will call die() and exit program so
the handle is leave not closed.
Regards,
ch3cooli
[-- Attachment #2: 0001-Also-close-config-file-handle-when-leaving-git_confi.patch --]
[-- Type: application/octet-stream, Size: 872 bytes --]
From 5d8d28f039b4db7b4f040a40fdc2edf866200f0c Mon Sep 17 00:00:00 2001
From: Sup Yut Sum <ch3cooli@gmail.com>
Date: Sat, 1 Dec 2012 21:56:41 +0800
Subject: [PATCH] Also close config file handle when leaving
git_config_set_multivar_in_file() early
Signed-off-by: Sup Yut Sum <ch3cooli@gmail.com>
---
config.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/config.c b/config.c
index fff8a43..4b5c877 100644
--- a/config.c
+++ b/config.c
@@ -1424,6 +1424,7 @@ int git_config_set_multivar_in_file(const char *config_filename,
contents = xmmap(NULL, contents_sz, PROT_READ,
MAP_PRIVATE, in_fd, 0);
close(in_fd);
+ in_fd = -1;
if (store.seen == 0)
store.seen = 1;
@@ -1493,6 +1494,8 @@ out_free:
if (lock)
rollback_lock_file(lock);
free(filename_buf);
+ if (in_fd >= 0)
+ close(in_fd);
return ret;
write_err_out:
--
1.8.0.msysgit.0
^ permalink raw reply related
* Re: [RFC] remove/deprecate 'submodule init' and 'sync'
From: W. Trevor King @ 2012-12-01 12:48 UTC (permalink / raw)
To: Phil Hord
Cc: Jens Lehmann, Git, Heiko Voigt, Junio C Hamano, Jeff King,
Shawn Pearce, Nahor
In-Reply-To: <CABURp0qNBcFnxbvhn7PsKWLUOsTiK4u5vx-=6cG3JQHw9aUeHA@mail.gmail.com>
[-- Attachment #1: Type: text/plain, Size: 785 bytes --]
On Fri, Nov 30, 2012 at 06:52:22PM -0500, Phil Hord wrote:
> If I never 'submodule init' a submodule, it does not get visited by
> 'git submodule foreach', among others. I think some people use this
> behavior explicitly.
This is something I'll fix while working up a trial patch. Currently
cmd_update calls module_clone if the <submodule>/.git does not exist.
This should probably happen in each command (in a wrapper around
module_list?). It's possible that module_list itself would need some
work, but I haven't absorbed its implementation yet [1].
Trevor
[1]: I read Perl by sounding out each letter ;).
--
This email may be signed or encrypted with GnuPG (http://www.gnupg.org).
For more information, see http://en.wikipedia.org/wiki/Pretty_Good_Privacy
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]
^ permalink raw reply
* Re: [PATCH/RFC 2/5] compat/terminal: factor out echo-disabling
From: Erik Faye-Lund @ 2012-12-01 12:43 UTC (permalink / raw)
To: Johannes Schindelin; +Cc: git, msysgit, peff
In-Reply-To: <alpine.DEB.1.00.1211301858570.31987@s15462909.onlinehome-server.info>
On Fri, Nov 30, 2012 at 6:59 PM, Johannes Schindelin
<Johannes.Schindelin@gmx.de> wrote:
> Hi,
>
> On Tue, 13 Nov 2012, Erik Faye-Lund wrote:
>
>> By moving the echo-disabling code to a separate function, we can
>> implement OS-specific versions of it for non-POSIX platforms.
>>
>> Signed-off-by: Erik Faye-Lund <kusmabite@gmail.com>
>> ---
>> compat/terminal.c | 43 +++++++++++++++++++++++++------------------
>> 1 file changed, 25 insertions(+), 18 deletions(-)
>>
>> diff --git a/compat/terminal.c b/compat/terminal.c
>> index bbb038d..3217838 100644
>> --- a/compat/terminal.c
>> +++ b/compat/terminal.c
>> @@ -14,6 +14,7 @@ static void restore_term(void)
>> return;
>>
>> tcsetattr(term_fd, TCSAFLUSH, &old_term);
>> + close(term_fd);
>> term_fd = -1;
>> }
>
> That looks like an independent resource leak fix... correct?
It might look like it, but it's not; term_fd used to be returned by
"fileno(fh)", and fh did get properly closed.
With my refactoring, disable_echo/restore_term takes opens /dev/tty a
second time, like Jeff points out. And that second file descriptor
needs to be closed.
--
*** Please reply-to-all at all times ***
*** (do not pretend to know who is subscribed and who is not) ***
*** Please avoid top-posting. ***
The msysGit Wiki is here: https://github.com/msysgit/msysgit/wiki - Github accounts are free.
You received this message because you are subscribed to the Google
Groups "msysGit" group.
To post to this group, send email to msysgit@googlegroups.com
To unsubscribe from this group, send email to
msysgit+unsubscribe@googlegroups.com
For more options, and view previous threads, visit this group at
http://groups.google.com/group/msysgit?hl=en_US?hl=en
^ permalink raw reply
* Re: [PATCH/RFC 1/5] mingw: make fgetc raise SIGINT if apropriate
From: Erik Faye-Lund @ 2012-12-01 12:36 UTC (permalink / raw)
To: Johannes Schindelin; +Cc: git, msysgit, peff
In-Reply-To: <alpine.DEB.1.00.1211301857170.31987@s15462909.onlinehome-server.info>
On Fri, Nov 30, 2012 at 6:58 PM, Johannes Schindelin
<Johannes.Schindelin@gmx.de> wrote:
> Hi,
>
> On Tue, 13 Nov 2012, Erik Faye-Lund wrote:
>
>> Set a control-handler to prevent the process from terminating, and
>> simulate SIGINT so it can be handled by a signal-handler as usual.
>
> One thing you might want to mention is that the fgetc() handling is not
> thread-safe, and intentionally so: if two threads read from the same
> console, we are in trouble anyway.
I'm not entirely sure if I know what you mean. Do you suggest that two
threads can race for setting the console ctrl-handler? I don't think
that's the case; "SetConsoleCtrlHandler(x, TRUE)" adds a console
handler to the handler-chain, and SetConsoleCtrlHandler(x, FALSE)
removes it. If two threads add handlers, it is my understanding that
one of them will be run, only to report "no, no more ctrl-handling
needed". Since both handlers block further ctrl-handling, I don't
think there's a problem.
Do you care to clarify what your thread-safety complaint is?
> BTW I like the new mingw_raise() very much!
Thanks! I originally implemented it for a different reason, but that
patch didn't turn out to be useful, so it's nice to finally put it to
use ;)
--
*** Please reply-to-all at all times ***
*** (do not pretend to know who is subscribed and who is not) ***
*** Please avoid top-posting. ***
The msysGit Wiki is here: https://github.com/msysgit/msysgit/wiki - Github accounts are free.
You received this message because you are subscribed to the Google
Groups "msysGit" group.
To post to this group, send email to msysgit@googlegroups.com
To unsubscribe from this group, send email to
msysgit+unsubscribe@googlegroups.com
For more options, and view previous threads, visit this group at
http://groups.google.com/group/msysgit?hl=en_US?hl=en
^ permalink raw reply
* Re: [PATCH 0/5] ignore SIG{INT,QUIT} when launching editor
From: Krzysztof Mazur @ 2012-12-01 12:34 UTC (permalink / raw)
To: Jeff King; +Cc: git, Junio C Hamano, Paul Fox
In-Reply-To: <20121130223943.GA27120@sigill.intra.peff.net>
On Fri, Nov 30, 2012 at 05:39:43PM -0500, Jeff King wrote:
> This is a re-roll of the pf/editor-ignore-sigint series.
>
> People mentioned some buggy editors which go into an infinite EIO loop
> when their parent dies due to SIGQUIT. That should be a non-issue now,
> as we will be ignoring SIGQUIT. And even if you could replicate it
> (e.g., with another signal) those programs should be (and reportedly
> have been) fixed. It is not git's job to babysit its child processes.
>
Also some good editors printed error message after they got EIO,
confusing the user.
Looks good to me. I've tested this with ed (always ignores SIGINT
and SIGQUIT), vim (always ignores SIGINT, but dies after three
SIGQUIT) and "sleep" (dies after SIGINT and SIGQUIT) and git works now
as expected. Doing what editor does is probably the best thing to do.
Tested-by: Krzysztof Mazur <krzysiek@podlesie.net>
Thanks,
Krzysiek
^ permalink raw reply
* Re: [PATCH/RFC 1/5] mingw: make fgetc raise SIGINT if apropriate
From: Erik Faye-Lund @ 2012-12-01 12:31 UTC (permalink / raw)
To: Jeff King; +Cc: Johannes Schindelin, git, msysgit
In-Reply-To: <20121130181119.GA7197@sigill.intra.peff.net>
On Fri, Nov 30, 2012 at 7:11 PM, Jeff King <peff@peff.net> wrote:
> On Fri, Nov 30, 2012 at 06:58:11PM +0100, Johannes Schindelin wrote:
>
>> Hi,
>>
>> On Tue, 13 Nov 2012, Erik Faye-Lund wrote:
>>
>> > Set a control-handler to prevent the process from terminating, and
>> > simulate SIGINT so it can be handled by a signal-handler as usual.
>>
>> One thing you might want to mention is that the fgetc() handling is not
>> thread-safe, and intentionally so: if two threads read from the same
>> console, we are in trouble anyway.
>
> That makes sense to me, but I'm confused why it is part of mingw_fgetc,
> which could in theory read from arbitrary streams, no? It it is not
> necessarily a console operation at all. I feel like I'm probably missing
> something subtle here...
I did add an early out for the non-console cases. Is this what you're
missing, perhaps?
--
*** Please reply-to-all at all times ***
*** (do not pretend to know who is subscribed and who is not) ***
*** Please avoid top-posting. ***
The msysGit Wiki is here: https://github.com/msysgit/msysgit/wiki - Github accounts are free.
You received this message because you are subscribed to the Google
Groups "msysGit" group.
To post to this group, send email to msysgit@googlegroups.com
To unsubscribe from this group, send email to
msysgit+unsubscribe@googlegroups.com
For more options, and view previous threads, visit this group at
http://groups.google.com/group/msysgit?hl=en_US?hl=en
^ permalink raw reply
* Bug report : gitattribute export-ignore behavior does not match documentation
From: Jean-Noël AVILA @ 2012-12-01 11:24 UTC (permalink / raw)
To: git
Tested on latest maint and master.
The 'export-ignore' gitattribute is documented as behaving on a pattern, just
like in .gitignore.
In repo where I have a tree like this:
.gitattributes
figures/
fr/figures/
I want to remove from archive all the figures directories. So I added
figures/ export-ignore
to the top level .gitattributes file.
but then:
$ git check-attr -a fr/figures
$ git check-attr -a figures
$
If I change the line in .gitattributes to :
figures export-ignore
I get the correct result, but I am not sure to exclude only directories.
Am I wrong somewhere?
Thank you.
Jean-Noël AVILA
^ permalink raw reply
* [PATCH v2 3/4] git-svn: Expand documentation for --follow-parent
From: Sebastian Leske @ 2012-11-30 7:16 UTC (permalink / raw)
To: git; +Cc: Michael J Gruber, Eric Wong
In-Reply-To: <cover.1354324110.git.Sebastian.Leske@sleske.name>
Describe what the option --follow-parent does, and what happens if it is
set or unset.
Signed-off-by: Sebastian Leske <sebastian.leske@sleske.name>
---
Documentation/git-svn.txt | 15 ++++++++++++---
1 file changed, 12 insertions(+), 3 deletions(-)
diff --git a/Documentation/git-svn.txt b/Documentation/git-svn.txt
index bfa8788..6bda014 100644
--- a/Documentation/git-svn.txt
+++ b/Documentation/git-svn.txt
@@ -628,10 +628,19 @@ ADVANCED OPTIONS
Default: "svn"
--follow-parent::
+ This option is only relevant if we are tracking branches (using
+ one of the repository layout options --trunk, --tags,
+ --branches, --stdlayout). For each tracked branch, try to find
+ out where its revision was copied (i.e. branched) from, and set
+ a suitable parent in the first git commit for the branch.
This is especially helpful when we're tracking a directory
- that has been moved around within the repository, or if we
- started tracking a branch and never tracked the trunk it was
- descended from. This feature is enabled by default, use
+ that has been moved around within the repository. If this
+ feature is disabled, the branches created by 'git svn' will all
+ be linear and not share any history, meaning that there will be
+ no information on where branches where branched off or merged.
+ However, following long/convoluted histories can take a long
+ time, so disabling this feature may speed up the cloning
+ process. This feature is enabled by default, use
--no-follow-parent to disable it.
+
[verse]
--
1.7.10.4
^ permalink raw reply related
* [PATCH v2 4/4] git-svn: Note about tags.
From: Sebastian Leske @ 2012-11-30 7:16 UTC (permalink / raw)
To: git; +Cc: Michael J Gruber, Eric Wong
In-Reply-To: <cover.1354324110.git.Sebastian.Leske@sleske.name>
Document that 'git svn' will import SVN tags as branches.
Signed-off-by: Sebastian Leske <sebastian.leske@sleske.name>
---
Documentation/git-svn.txt | 7 +++++++
1 file changed, 7 insertions(+)
diff --git a/Documentation/git-svn.txt b/Documentation/git-svn.txt
index 6bda014..18d5e45 100644
--- a/Documentation/git-svn.txt
+++ b/Documentation/git-svn.txt
@@ -959,6 +959,13 @@ the possible corner cases (git doesn't do it, either). Committing
renamed and copied files is fully supported if they're similar enough
for git to detect them.
+SVN tags (if tracked using options '--tags' or '--stdlayout') are
+imported as git branches, prefixing the tag name with 'tags/'.
+This is because tags in SVN behave more like git branches: The contents
+of a tag need not be identical to the tagged commit, and it is possible
+(though discouraged) to commit changes to a tag (because a tag is just a
+directory copy, thus technically the same as a branch).
+
CONFIGURATION
-------------
--
1.7.10.4
^ permalink raw reply related
* [PATCH v2 2/4] Recommend use of structure options for git svn.
From: Sebastian Leske @ 2012-11-30 7:16 UTC (permalink / raw)
To: git; +Cc: Michael J Gruber, Eric Wong
In-Reply-To: <cover.1354324110.git.Sebastian.Leske@sleske.name>
Document that when using git svn, one should usually either use the
directory structure options to import branches as branches, or only
import one subdirectory. The default behaviour of cloning all branches
and tags as subdirectories in the working copy is usually not what the
user wants.
Signed-off-by: Sebastian Leske <sebastian.leske@sleske.name>
---
Documentation/git-svn.txt | 24 +++++++++++++++++++++---
1 file changed, 21 insertions(+), 3 deletions(-)
diff --git a/Documentation/git-svn.txt b/Documentation/git-svn.txt
index 824bf82..bfa8788 100644
--- a/Documentation/git-svn.txt
+++ b/Documentation/git-svn.txt
@@ -739,7 +739,8 @@ for rewriteRoot and rewriteUUID which can be used together.
BASIC EXAMPLES
--------------
-Tracking and contributing to the trunk of a Subversion-managed project:
+Tracking and contributing to the trunk of a Subversion-managed project
+(ignoring tags and branches):
------------------------------------------------------------------------
# Clone a repo (like git clone):
@@ -764,8 +765,10 @@ Tracking and contributing to an entire Subversion-managed project
(complete with a trunk, tags and branches):
------------------------------------------------------------------------
-# Clone a repo (like git clone):
- git svn clone http://svn.example.com/project -T trunk -b branches -t tags
+# Clone a repo with standard SVN directory layout (like git clone):
+ git svn clone http://svn.example.com/project --stdlayout
+# Or, if the repo uses a non-standard directory layout:
+ git svn clone http://svn.example.com/project -T tr -b branch -t tag
# View all branches and tags you have cloned:
git branch -r
# Create a new branch in SVN
@@ -871,6 +874,21 @@ already dcommitted. It is considered bad practice to --amend commits
you've already pushed to a remote repository for other users, and
dcommit with SVN is analogous to that.
+When cloning an SVN repository, if none of the options for describing
+the repository layout is used (--trunk, --tags, --branches,
+--stdlayout), 'git svn clone' will create a git repository with
+completely linear history, where branches and tags appear as separate
+folders in the working copy. While this is the easiest way to get a
+copy of a complete repository, for projects with many branches it will
+lead to a working copy many times larger than just the trunk. Thus for
+projects using the standard directory structure (trunk/branches/tags),
+it is recommended to clone with option '--stdlayout'. If the project
+uses a non-standard structure, and/or if branches and tags are not
+required, it is easiest to only clone one directory (typically trunk),
+without giving any repository layout options. If the full history with
+branches and tags is required, the options '--trunk' / '--branches' /
+'--tags' must be used.
+
When using multiple --branches or --tags, 'git svn' does not automatically
handle name collisions (for example, if two branches from different paths have
the same name, or if a branch and a tag have the same name). In these cases,
--
1.7.10.4
^ permalink raw reply related
* [PATCH v2 0/4] git-svn: More docs for branch handling in
From: Sebastian Leske @ 2012-12-01 1:08 UTC (permalink / raw)
To: git; +Cc: Michael J Gruber, Eric Wong
Updated version of my documentation patch for git-svn. Thanks to Michael
J Gruber and Eric Wong for helpful comments.
Sebastian Leske (4):
git-svn: Document branches with at-sign(@).
Recommend use of structure options for git svn.
git-svn: Expand documentation for --follow-parent
git-svn: Note about tags.
Documentation/git-svn.txt | 84 +++++++++++++++++++++++++++++++++++++++++----
1 file changed, 78 insertions(+), 6 deletions(-)
--
1.7.10.4
^ permalink raw reply
* [PATCH v2 1/4] git-svn: Document branches with at-sign(@).
From: Sebastian Leske @ 2012-11-30 7:16 UTC (permalink / raw)
To: git; +Cc: Michael J Gruber, Eric Wong
In-Reply-To: <cover.1354324110.git.Sebastian.Leske@sleske.name>
git svn will sometimes create branches with an at-sign in the name
(branchname@revision). These branches confuse many users and it is a FAQ
why they are created. Document when git svn will create them.
Signed-off-by: Sebastian Leske <sebastian.leske@sleske.name>
---
Documentation/git-svn.txt | 38 ++++++++++++++++++++++++++++++++++++++
1 file changed, 38 insertions(+)
diff --git a/Documentation/git-svn.txt b/Documentation/git-svn.txt
index 8b0d3ad..824bf82 100644
--- a/Documentation/git-svn.txt
+++ b/Documentation/git-svn.txt
@@ -881,6 +881,44 @@ different name spaces. For example:
branches = stable/*:refs/remotes/svn/stable/*
branches = debug/*:refs/remotes/svn/debug/*
+If 'git svn' is configured to fetch branches (and --follow-branches
+is in effect), it will sometimes create multiple branches for one SVN
+branch, where the addtional branches have names of the form
+'branchname@nnn' (with nnn an SVN revision number). These additional
+branches are created if 'git svn' cannot find a parent commit for the
+first commit in an SVN branch, to connect the branch to the history of
+the other branches. Normally, the first commit in an SVN branch consists
+of a copy operation. 'git svn' will read this commit to get the SVN
+revision the branch was created (copied) from. It will then try to find the
+git commit that corresponds to this SVN revision, and use that as the
+parent of the branch. However, it is possible that there is no suitable
+git commit to serve as parent. This will happen, among other reasons,
+if the SVN branch is a copy of a revision that was not fetched by 'git
+svn' (e.g. because it is an old revision that was skipped with
+'--revision'), or if in SVN a directory was copied that is not tracked
+by 'git svn' (such as a branch that is not tracked at all, or a
+subdirectory of a tracked branch). In these cases, 'git svn' will still
+create a git branch, but instead of using an existing git commit as the
+parent of the branch, it will read the SVN history of the directory the
+branch was copied from and create appropriate git commits (this is
+indicated by the message "Initializing parent: <branchname>").
+Additionally, it will create a special branch named
+'<branchname>@<SVN-Revision>', where <SVN-Revision> is the SVN revision
+number the branch was copied from. This branch will point to the newly
+created parent commit of the branch. If in SVN the branch was deleted
+and later recreated from a different version, there will be multiple
+such branches with an '@'.
+Note that this may mean that multiple git commits are created for a
+single SVN revision. An example: In an SVN repository with a standard
+trunk/tags/branches layout, a directory trunk/sub is created in r.100.
+In r.200, trunk/sub is branched by copying it to branches/. 'git svn
+clone -s' will then create a branch 'sub'. It will also create new git
+commits for r.100 through r.199 and use these as the history of branch
+'sub'. Thus there will be two git commits for each revision from r.100
+to r.199 (one containing trunk/, one containing trunk/sub/). Finally,
+it will create a branch 'sub@200' pointing to the new parent commit of
+branch 'sub' (i.e. the commit for r.200 and trunk/sub/).
+
BUGS
----
--
1.7.10.4
^ permalink raw reply related
* Re: [PATCH 0/5] "diff --stat" counting fixes
From: Antoine Pelisse @ 2012-12-01 10:29 UTC (permalink / raw)
To: Junio C Hamano; +Cc: git
In-Reply-To: <1354051310-29093-1-git-send-email-gitster@pobox.com>
Hi Junio,
That does make a lot of sense and I would have indeed missed a couple
of things here.
I've been thinking about that "Unmerged" line quite a lot, and I can't
get myself any good reason to keep it.
Would you mind taking a couple of minutes to make it clear ?
I feel like (but I can obviously be wrong):
1. The info is redundant. When performing a merge, all diffs (without
--staged flag) are unmerged
2. While status shows the line once, while diff shows the diff for the file
once, while diff --shortstat counts the file once, diff --stat shows two
lines for the file.
3. diff --numstat shows two lines for the same file. As a script
writer (I guess that's what it's meant for), I would definitely expect
uniqueness in third column/filenames.
Cheers,
Antoine
On Tue, Nov 27, 2012 at 10:21 PM, Junio C Hamano <gitster@pobox.com> wrote:
>
> It turns out that there are at least two bugs in the diffstat
> counting code. This series comes on top of the earlier 74faaa1 (Fix
> "git diff --stat" for interesting - but empty - file changes,
> 2012-10-17) to fix them.
>
> Junio C Hamano (5):
> test: add failing tests for "diff --stat" to t4049
> diff --stat: status of unmodified pair in diff-q is not zero
> diff --stat: use "file" temporary variable to refer to data->files[i]
> diff --stat: move the "total count" logic to the last loop
> diff --stat: do not count "unmerged" entries
>
> diff.c | 49 +++++++++++++++++++++++++---------------------
> t/t4049-diff-stat-count.sh | 46 ++++++++++++++++++++++++++++++++++++++++++-
> 2 files changed, 72 insertions(+), 23 deletions(-)
>
> --
> 1.8.0.1.331.g808d2af
>
^ permalink raw reply
* Re: [PATCH] fast-export: Allow pruned-references in mark file
From: Antoine Pelisse @ 2012-12-01 10:10 UTC (permalink / raw)
To: Junio C Hamano; +Cc: Felipe Contreras, git
In-Reply-To: <7vobikqelo.fsf@alter.siamese.dyndns.org>
> Yeah, I think I agree that you would need to make sure that the
> other side does not use the revision marked with :2, once you retire
> the object you originally marked with :2 by pruning. Shouldn't the
> second export show :1 and :3 but not :2? It feels like a bug in the
> exporter to me that the mark number is reused in such a case.
It depends what you call a bug.
If the last item from the list is pruned, and no new objects
are exported, you will lose both reference and count to mark :2.
In this situation, incrementing last_idnum was pointless.
Assuming that we can't do anything about that, marks should be
considered mutable (and I don't think there is any way it
shouldn't). Then incrementing last_idnum is always useless.
Now, if marks can change, I don't understand why we use them at all.
(or don't provide the possibility to not use them at least).
In the "hg <-> git" case, it seems like an unecessary step:
hg revs <-> git marks <-> git sha1
Potentially forces the remote-helper to re-read the "marks <-> sha1"
everytime.
Also in the remote-helper, the "list" command requires sha1 for each
heads, while "import/export" can't work with sha1 but only marks, which
seems inconsistent.
My last point is about "git-remote-hg" and still mutable revs.
It seems like Felipe is using revs() rather than node() or hex() to
refer to mercurial changeset while those revs are also mutable, and
there exists immutable references: hex.
To sum up, the whole idea is, why would we use unsafe mutable marks
when we can use safer immutable references ?
Cheers,
Antoine
^ permalink raw reply
* Re: [RFC/PATCH 1/2] reset: learn to reset to tree
From: Junio C Hamano @ 2012-12-01 9:24 UTC (permalink / raw)
To: Martin von Zweigbergk; +Cc: git
In-Reply-To: <CANiSa6i2f-4jXFUpYV6+fYnpG-tSRRA3cRg_v-v=UrgfwfFz_g@mail.gmail.com>
Martin von Zweigbergk <martinvonz@gmail.com> writes:
> On Thu, Nov 29, 2012 at 2:00 PM, Martin von Zweigbergk
> <martinvonz@gmail.com> wrote:
>> Slightly off topic, but another difference (or somehow another aspect
>> of the same difference?) that has tripped me up a few times is that
>> "git checkout $rev ." only affects added and modified files...
"checkout $commit pathspec" has always been about checking out the
contents stored in the paths that match the pathspec from the named
commit to the index and also o the working tree. "checkout
pathspec" is similar, but the stuff comes out of the index.
When pathspec is "dir/", it does not match the directory whose name
is "dir". The pathspec matches the paths that store blobs under
that directory.
In other words, "checkout dir/" (or "checkout HEAD~4 dir/) has never
been about "please remove everything in dir/, and then check out
everything in dir/ from the index (or from HEAD~4)". The "please
remove everything in dir/" part is not the job of "checkout"; of
course, you can do it as a separate step (e.g. "rm -fr dir/").
^ permalink raw reply
* Stitching histories of several repositories
From: Ramkumar Ramachandra @ 2012-12-01 9:11 UTC (permalink / raw)
To: Git List
Hi,
I've written a tool to stitch the first-parent histories of several
git repositories. To illustrate, consider that we have a toplevel git
repository inside which the other repositories reside.
/.git
/a/.git
/b/.git
The tool pulls in the objects from /a/.git and /b/.git into /.git by
using alternates. Then, it gets the list of commits by rev-list'ing
and emits a fast-import stream rewriting the paths of the files in a/
from / to a/ and the paths of files in b/ form / to b/ (using M
040000), taking care to have one unified first-parent line. It turns
out that this is quite non-trivial when the individual git
repositories contain lots of merges. To illustrate, this is an
example of how histories from two repositories with one merge commit
each would be stitched:
* c1f81f1 (refs/figlets/first/origin/master)
|\
| * 6eaf22f (refs/figlets/first-2/origin/master)
* 15cd841
* 1f2f408 (refs/figlets/second/origin/master)
|\
| * 4bd7fe3 (refs/figlets/second-2/origin/master)
* bfa9f3c
* 5e1d337 (refs/replay/stitch2, fig/stitch2)
|\
| * 23437d5 (refs/replay/1f2f40-2)
|/
* 4f2d70b
|\
| * e017fa5 (refs/replay/c1f81f-2)
* c0251c7
* 4036fea
I'd like to know whether the tool would be useful to a wider audience,
before I polish it and consider submitting it for inclusion in
contrib/. I think the tool is especially useful for running bisect
and tracking bugs that occur in large projects that consist of many
git repositories. Will a unified log showing commits in different
submodules be useful?
Ram
^ permalink raw reply
* Re: Topics currently in the Stalled category
From: Adam Spiers @ 2012-12-01 0:36 UTC (permalink / raw)
To: Junio C Hamano; +Cc: git
In-Reply-To: <7vobirq0q2.fsf_-_@alter.siamese.dyndns.org>
On Wed, Nov 21, 2012 at 12:05 AM, Junio C Hamano <gitster@pobox.com> wrote:
> Here is a list of stalled topics I am having trouble deciding what
> to do (the default is to dismiss them around feature freeze).
[snipped]
> * as/check-ignore (2012-11-08) 14 commits
> - t0007: fix tests on Windows
> - Documentation/check-ignore: we show the deciding match, not the first
> - Add git-check-ignore sub-command
> - dir.c: provide free_directory() for reclaiming dir_struct memory
> - pathspec.c: move reusable code from builtin/add.c
> - dir.c: refactor treat_gitlinks()
> - dir.c: keep track of where patterns came from
> - dir.c: refactor is_path_excluded()
> - dir.c: refactor is_excluded()
> - dir.c: refactor is_excluded_from_list()
> - dir.c: rename excluded() to is_excluded()
> - dir.c: rename excluded_from_list() to is_excluded_from_list()
> - dir.c: rename path_excluded() to is_path_excluded()
> - dir.c: rename cryptic 'which' variable to more consistent name
>
> Duy helped to reroll this, but it seems that there weren't any
> activity since then during my absense.
I have been delayed several times, but I finally resumed work on
another re-roll. I don't think there is any major reworking required;
just a number of small tweaks.
> * as/test-tweaks (2012-09-20) 7 commits
> - tests: paint unexpectedly fixed known breakages in bold red
> - tests: test the test framework more thoroughly
> - [SQUASH] t/t0000-basic.sh: quoting of TEST_DIRECTORY is screwed up
> - tests: refactor mechanics of testing in a sub test-lib
> - tests: paint skipped tests in bold blue
> - tests: test number comes first in 'not ok $count - $message'
> - tests: paint known breakages in bold yellow
>
> Various minor tweaks to the test framework to paint its output
> lines in colors that match what they mean better.
>
> Has the "is this really blue?" issue Peff raised resolved???
I have a re-roll of this ready - just need to rebase to latest master,
do a final sanity check, and then send.
Sorry again for the delays.
^ permalink raw reply
* Re: [RFC] remove/deprecate 'submodule init' and 'sync'
From: Phil Hord @ 2012-11-30 23:52 UTC (permalink / raw)
To: W. Trevor King
Cc: Jens Lehmann, Git, Heiko Voigt, Junio C Hamano, Jeff King,
Shawn Pearce, Nahor
In-Reply-To: <20121130175309.GA718@odin.tremily.us>
On Fri, Nov 30, 2012 at 12:53 PM, W. Trevor King <wking@tremily.us> wrote:
> On Wed, Nov 28, 2012 at 12:19:04AM +0100, Jens Lehmann wrote:
>> Am 26.11.2012 22:00, schrieb W. Trevor King:
>> > From: "W. Trevor King" <wking@tremily.us>
>> >
>> > This allows users to override the .gitmodules value with a
>> > per-repository value.
>>
>> Your intentions makes lots of sense, but your patch does more than
>> that. Copying the branch setting into .git/config sets the initial
>> branch setting into stone. That makes it impossible to have a branch
>> "foo" in the superproject using a branch "bar" in a submodule and
>> another superproject branch "frotz" using branch "nitfol" for the
>> same submodule. You should use the branch setting from .git/config
>> if present and fall back to the branch setting from .gitmodules if
>> not, which would enable the user to have her own setting if she
>> doesn't like what upstream provides but would still enable others
>> to follow different submodule branches in different superproject
>> branches.
>
> I've mulling this over, and when I started coding support for
> submodule.<name>.remote, I had an idea.
>
> On Thu, Nov 29, 2012 at 10:27:19PM -0500, W. Trevor King wrote:
>> On Thu, Nov 29, 2012 at 08:11:20PM -0500, Phil Hord wrote:
>> > I've always felt that the "origin" defaults are broken and are simply
>> > being ignored because most users do not trip over them. But ISTR that
>> > submodule commands use the remote indicated by the superproject's
>> > current remote-tracking configuration, with a fallback to 'origin' if
>> > there is none. Sort of a "best effort" algorithm, I think. Am I
>> > remembering that wrong?
>>
>> The current code uses a bare "git-fetch". I'm not sure what that
>> defaults to if you're on a detached head. If it bothers you, I'm fine
>> adding the submodule.<name>.remote option in v6.
>
> In my v5 patch, I check for submodule.<name>.remote first in the usual
> `git config` files. If I don't find what I'm looking for I fall back
> on .gitmodules (basically Jens' suggestion). However, my initial
> copying-to-.git/config approach was mostly done to mimic existing
> configuration handling in git-submodule.sh. Since I agree with Jens
> on configuration precendence, and I now had two options to read
> (.branch and .remote), I thought I'd pull the logic out into its own
> function (code included at the end). While I was shifting the
> existing submodule config handling over to my new function, I noticed
> that with this logic, `submodule init` doesn't really do anything
> important anymore.
If I never 'submodule init' a submodule, it does not get visited by
'git submodule foreach', among others. I think some people use this
behavior explicitly.
On the other hand, I've also notice that a submodule which I have
removed does not get de-inited later one. It causes my 'git submodule
foreach' to emit errors. :-(
Phil
> Likewise for `submodule sync`, which seems to be
> quite similar to `init`.
>
> What to do about this? `init` has been around for a while, so we
> can't just remove it (maybe in 2.0?). Leaving it in place is not
> really a problem though, it just means that the user is locking in the
> current .gitmodules configuration (as Jens pointed out with respect to
> .branch).
>
> I may be way off base here, as I'm fairly new to submodules in general
> and these two commands in particular, but I thought I'd float the
> idea.
>
> Cheers,
> Trevor
>
> ---
> #
> # Print a submodule configuration setting
> #
> # $1 = submodule name
> # $2 = option name
> # $3 = default value
> #
> # Checks in the usual git-config places first (for overrides),
> # otherwise it falls back on .gitmodules. This allows you to
> # distribute project-wide defaults in .gitmodules, while still
> # customizing individual repositories if necessary. If the option is
> # not in .gitmodules either, print a default value.
> #
> get_submodule_config()
> {
> name="$1"
> option="$2"
> default="$3"
> value=$(git config submodule."$name"."$option")
> if test -z "$value"
> then
> value=$(git config -f .gitmodules submodule."$name"."$option")
> fi
> printf '%s' "${value:-$default}"
> }
>
> --
> This email may be signed or encrypted with GnuPG (http://www.gnupg.org).
> For more information, see http://en.wikipedia.org/wiki/Pretty_Good_Privacy
^ permalink raw reply
* [PATCH 5/5] launch_editor: propagate signals from editor to git
From: Jeff King @ 2012-11-30 22:41 UTC (permalink / raw)
To: git; +Cc: Junio C Hamano, Paul Fox, Krzysztof Mazur
In-Reply-To: <20121130223943.GA27120@sigill.intra.peff.net>
We block SIGINT and SIGQUIT while the editor runs so that
git is not killed accidentally by a stray "^C" meant for the
editor or its subprocesses. This works because most editors
ignore SIGINT.
However, some editor wrappers, like emacsclient, expect to
die due to ^C. We detect the signal death in the editor and
properly exit, but not before writing a useless error
message to stderr. Instead, let's notice when the editor was
killed by a terminal signal and just raise the signal on
ourselves. This skips the message and looks to our parent
like we received SIGINT ourselves.
The end effect is that if the user's editor ignores SIGINT,
we will, too. And if it does not, then we will behave as if
we did not ignore it. That should make all users happy.
Note that in the off chance that another part of git has
ignored SIGINT while calling launch_editor, we will still
properly detect and propagate the failed return code from
the editor (i.e., the worst case is that we generate the
useless error, not fail to notice the editor's death).
Signed-off-by: Jeff King <peff@peff.net>
---
editor.c | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/editor.c b/editor.c
index c892a81..065a7ab 100644
--- a/editor.c
+++ b/editor.c
@@ -39,7 +39,7 @@ int launch_editor(const char *path, struct strbuf *buffer, const char *const *en
if (strcmp(editor, ":")) {
const char *args[] = { editor, path, NULL };
struct child_process p;
- int ret;
+ int ret, sig;
memset(&p, 0, sizeof(p));
p.argv = args;
@@ -51,8 +51,11 @@ int launch_editor(const char *path, struct strbuf *buffer, const char *const *en
sigchain_push(SIGINT, SIG_IGN);
sigchain_push(SIGQUIT, SIG_IGN);
ret = finish_command(&p);
+ sig = ret + 128;
sigchain_pop(SIGINT);
sigchain_pop(SIGQUIT);
+ if (sig == SIGINT || sig == SIGQUIT)
+ raise(sig);
if (ret)
return error("There was a problem with the editor '%s'.",
editor);
--
1.8.0.1.620.g558b0aa
^ permalink raw reply related
* [PATCH 4/5] run-command: do not warn about child death from terminal
From: Jeff King @ 2012-11-30 22:41 UTC (permalink / raw)
To: git; +Cc: Junio C Hamano, Paul Fox, Krzysztof Mazur
In-Reply-To: <20121130223943.GA27120@sigill.intra.peff.net>
SIGINT and SIGQUIT are not generally interesting signals to
the user, since they are typically caused by them hitting "^C"
or otherwise telling their terminal to send the signal.
Signed-off-by: Jeff King <peff@peff.net>
---
run-command.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/run-command.c b/run-command.c
index 3aae270..757f263 100644
--- a/run-command.c
+++ b/run-command.c
@@ -242,7 +242,8 @@ static int wait_or_whine(pid_t pid, const char *argv0)
error("waitpid is confused (%s)", argv0);
} else if (WIFSIGNALED(status)) {
code = WTERMSIG(status);
- error("%s died of signal %d", argv0, code);
+ if (code != SIGINT && code != SIGQUIT)
+ error("%s died of signal %d", argv0, code);
/*
* This return value is chosen so that code & 0xff
* mimics the exit code that a POSIX shell would report for
--
1.8.0.1.620.g558b0aa
^ permalink raw reply related
* [PATCH 3/5] launch_editor: ignore terminal signals while editor has control
From: Jeff King @ 2012-11-30 22:41 UTC (permalink / raw)
To: git; +Cc: Junio C Hamano, Paul Fox, Krzysztof Mazur
In-Reply-To: <20121130223943.GA27120@sigill.intra.peff.net>
From: Paul Fox <pgf@foxharp.boston.ma.us>
The user's editor likely catches SIGINT (ctrl-C). but if
the user spawns a command from the editor and uses ctrl-C to
kill that command, the SIGINT will likely also kill git
itself (depending on the editor, this can leave the terminal
in an unusable state).
Let's ignore it while the editor is running, and do the same
for SIGQUIT, which many editors also ignore. This matches
the behavior if we were to use system(3) instead of
run-command.
Signed-off-by: Paul Fox <pgf@foxharp.boston.ma.us>
Signed-off-by: Jeff King <peff@peff.net>
---
editor.c | 9 ++++++++-
1 file changed, 8 insertions(+), 1 deletion(-)
diff --git a/editor.c b/editor.c
index 842f782..c892a81 100644
--- a/editor.c
+++ b/editor.c
@@ -1,6 +1,7 @@
#include "cache.h"
#include "strbuf.h"
#include "run-command.h"
+#include "sigchain.h"
#ifndef DEFAULT_EDITOR
#define DEFAULT_EDITOR "vi"
@@ -38,6 +39,7 @@ int launch_editor(const char *path, struct strbuf *buffer, const char *const *en
if (strcmp(editor, ":")) {
const char *args[] = { editor, path, NULL };
struct child_process p;
+ int ret;
memset(&p, 0, sizeof(p));
p.argv = args;
@@ -46,7 +48,12 @@ int launch_editor(const char *path, struct strbuf *buffer, const char *const *en
if (start_command(&p) < 0)
return error("unable to start editor '%s'", editor);
- if (finish_command(&p))
+ sigchain_push(SIGINT, SIG_IGN);
+ sigchain_push(SIGQUIT, SIG_IGN);
+ ret = finish_command(&p);
+ sigchain_pop(SIGINT);
+ sigchain_pop(SIGQUIT);
+ if (ret)
return error("There was a problem with the editor '%s'.",
editor);
}
--
1.8.0.1.620.g558b0aa
^ permalink raw reply related
* [PATCH 2/5] launch_editor: refactor to use start/finish_command
From: Jeff King @ 2012-11-30 22:41 UTC (permalink / raw)
To: git; +Cc: Junio C Hamano, Paul Fox, Krzysztof Mazur
In-Reply-To: <20121130223943.GA27120@sigill.intra.peff.net>
The launch_editor function uses the convenient run_command_*
interface. Let's use the more flexible start_command and
finish_command functions, which will let us manipulate the
parent state while we're waiting for the child to finish.
Signed-off-by: Jeff King <peff@peff.net>
---
editor.c | 10 +++++++++-
1 file changed, 9 insertions(+), 1 deletion(-)
diff --git a/editor.c b/editor.c
index d834003..842f782 100644
--- a/editor.c
+++ b/editor.c
@@ -37,8 +37,16 @@ int launch_editor(const char *path, struct strbuf *buffer, const char *const *en
if (strcmp(editor, ":")) {
const char *args[] = { editor, path, NULL };
+ struct child_process p;
- if (run_command_v_opt_cd_env(args, RUN_USING_SHELL, NULL, env))
+ memset(&p, 0, sizeof(p));
+ p.argv = args;
+ p.env = env;
+ p.use_shell = 1;
+ if (start_command(&p) < 0)
+ return error("unable to start editor '%s'", editor);
+
+ if (finish_command(&p))
return error("There was a problem with the editor '%s'.",
editor);
}
--
1.8.0.1.620.g558b0aa
^ permalink raw reply related
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