* linux-2.6.git/packed-refs???
@ 2007-02-13 9:27 Junio C Hamano
2007-02-13 9:48 ` [PATCH] Do not run git-pack-refs by default for now from git-gc Junio C Hamano
2007-02-13 18:16 ` linux-2.6.git/packed-refs??? Theodore Tso
0 siblings, 2 replies; 9+ messages in thread
From: Junio C Hamano @ 2007-02-13 9:27 UTC (permalink / raw)
To: torvalds; +Cc: git
I noticed while scanning #git log that it appears that pack-refs
was run in your public repo and some people cannot clone over
dumb protocols with older git.
commit 2986c02217f98809d8990e7679edf0f5d99f904d
Author: Junio C Hamano <junkio@cox.net>
Date: Wed Nov 22 22:24:09 2006 -0800
git-fetch: fix dumb protocol transport to fetch from pack-pruned ref
was the first revision that aligned dumb protocol clients with
pack-ref; unfortunately anything older will not find the ref
that was packed.
We do have a backward compatibility warning on this in v1.5.0
release notes draft, but I think we would make it more obvious.
Sigh...
^ permalink raw reply [flat|nested] 9+ messages in thread
* [PATCH] Do not run git-pack-refs by default for now from git-gc
2007-02-13 9:27 linux-2.6.git/packed-refs??? Junio C Hamano
@ 2007-02-13 9:48 ` Junio C Hamano
2007-02-13 10:03 ` Johannes Schindelin
2007-02-13 18:16 ` linux-2.6.git/packed-refs??? Theodore Tso
1 sibling, 1 reply; 9+ messages in thread
From: Junio C Hamano @ 2007-02-13 9:48 UTC (permalink / raw)
To: torvalds; +Cc: git
Just to be on the safe side, let's not run pack-refs from git-gc
without an explicit "[gc] packrefs = true" configuration. This
default should be changed in the future when we can reasonably
assume that everybody runs post v1.5.0-rc0 version of git, at
which time people can still ask it not to run with the same
configuration ("[gc] packrefs = false").
Signed-off-by: Junio C Hamano <junkio@cox.net>
---
diff --git a/Documentation/config.txt b/Documentation/config.txt
index 0129b1f..36b2c4f 100644
--- a/Documentation/config.txt
+++ b/Documentation/config.txt
@@ -321,6 +321,16 @@ format.headers::
Additional email headers to include in a patch to be submitted
by mail. See gitlink:git-format-patch[1].
+gc.packrefs::
+ `git gc` does not run `git pack-refs` by default
+ so that older dumb-transport clients can still fetch
+ from the repository. Setting this to `true` lets `git
+ gc` to run `git pack-refs`. Enable it only when you
+ know you do not have to support such clients. The
+ default will change to run `git pack-refs` in some
+ future, and setting this to `false` will continue to
+ prevent `git pack-refs` from being run from `git gc`.
+
gc.reflogexpire::
`git reflog expire` removes reflog entries older than
this time; defaults to 90 days.
diff --git a/Documentation/git-gc.txt b/Documentation/git-gc.txt
index e37758a..ca38c2e 100644
--- a/Documentation/git-gc.txt
+++ b/Documentation/git-gc.txt
@@ -62,9 +62,14 @@ The optional configuration variable 'gc.rerereunresolved' indicates
how long records of conflicted merge you have not resolved are
kept. This defaults to 15 days.
+The optional configuration variable 'gc.packrefs' determines if
+`git gc` runs `git-pack-refs`. Without the configuration, `git-pack-refs`
+is not run by default to allow older dumb-transport clients
+fetch from the repository, but this will change in the future.
See Also
--------
+gitlink:git-pack-refs[1]
gitlink:git-prune[1]
gitlink:git-reflog[1]
gitlink:git-repack[1]
diff --git a/git-gc.sh b/git-gc.sh
index 3e8c87c..054f338 100755
--- a/git-gc.sh
+++ b/git-gc.sh
@@ -22,7 +22,11 @@ do
shift
done
-git-pack-refs --prune &&
+if pack_refs=$(git config --bool --get gc.packrefs) &&
+ test "true" = "$pack_refs"
+then
+ git-pack-refs --prune
+fi &&
git-reflog expire --all &&
git-repack -a -d -l &&
$no_prune git-prune &&
^ permalink raw reply related [flat|nested] 9+ messages in thread
* Re: [PATCH] Do not run git-pack-refs by default for now from git-gc
2007-02-13 9:48 ` [PATCH] Do not run git-pack-refs by default for now from git-gc Junio C Hamano
@ 2007-02-13 10:03 ` Johannes Schindelin
2007-02-13 10:29 ` Jeff King
0 siblings, 1 reply; 9+ messages in thread
From: Johannes Schindelin @ 2007-02-13 10:03 UTC (permalink / raw)
To: Junio C Hamano; +Cc: torvalds, git
Hi,
On Tue, 13 Feb 2007, Junio C Hamano wrote:
> Just to be on the safe side, let's not run pack-refs from git-gc
> without an explicit "[gc] packrefs = true" configuration.
It seems to me that the reason are dumb transports, which are very likely
to run only from bare repositories. How about checking for a bare
repository explicitely, and only if it _is_ bare, check for gc.packrefs,
too?
Packed refs _do_ have a tremendous advantage (when you have a lot of tags,
like many CVS switchers).
Ciao,
Dscho
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH] Do not run git-pack-refs by default for now from git-gc
2007-02-13 10:03 ` Johannes Schindelin
@ 2007-02-13 10:29 ` Jeff King
2007-02-13 10:55 ` Johannes Schindelin
2007-02-13 13:01 ` [PATCH] git-gc: run pack-refs by default unless the repo is bare Johannes Schindelin
0 siblings, 2 replies; 9+ messages in thread
From: Jeff King @ 2007-02-13 10:29 UTC (permalink / raw)
To: Johannes Schindelin; +Cc: git
On Tue, Feb 13, 2007 at 11:03:32AM +0100, Johannes Schindelin wrote:
> It seems to me that the reason are dumb transports, which are very likely
> to run only from bare repositories. How about checking for a bare
> repository explicitely, and only if it _is_ bare, check for gc.packrefs,
> too?
The way you have stated it, I think we will get a lot of "I set
gc.packrefs, but it doesn't do anything!" complaints. I think a
tri-state "yes/no/notbare" defaulting to "notbare" makes more sense. But
maybe you meant the other way around in the first place.
-Peff
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH] Do not run git-pack-refs by default for now from git-gc
2007-02-13 10:29 ` Jeff King
@ 2007-02-13 10:55 ` Johannes Schindelin
2007-02-13 17:59 ` Junio C Hamano
2007-02-13 13:01 ` [PATCH] git-gc: run pack-refs by default unless the repo is bare Johannes Schindelin
1 sibling, 1 reply; 9+ messages in thread
From: Johannes Schindelin @ 2007-02-13 10:55 UTC (permalink / raw)
To: Jeff King; +Cc: git
Hi,
On Tue, 13 Feb 2007, Jeff King wrote:
> On Tue, Feb 13, 2007 at 11:03:32AM +0100, Johannes Schindelin wrote:
>
> > It seems to me that the reason are dumb transports, which are very
> > likely to run only from bare repositories. How about checking for a
> > bare repository explicitely, and only if it _is_ bare, check for
> > gc.packrefs, too?
>
> The way you have stated it, I think we will get a lot of "I set
> gc.packrefs, but it doesn't do anything!" complaints. I think a
> tri-state "yes/no/notbare" defaulting to "notbare" makes more sense. But
> maybe you meant the other way around in the first place.
No, I meant it as you understood it. But your solution is evidently way
more elegant.
Ciao,
Dscho
^ permalink raw reply [flat|nested] 9+ messages in thread
* [PATCH] git-gc: run pack-refs by default unless the repo is bare
2007-02-13 10:29 ` Jeff King
2007-02-13 10:55 ` Johannes Schindelin
@ 2007-02-13 13:01 ` Johannes Schindelin
1 sibling, 0 replies; 9+ messages in thread
From: Johannes Schindelin @ 2007-02-13 13:01 UTC (permalink / raw)
To: Jeff King, junkio; +Cc: git
The config variable gc.packrefs is tristate now: "true", "false"
and "notbare", where "notbare" is the default.
Signed-off-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
---
On Tue, 13 Feb 2007, Jeff King wrote:
> I think a tri-state "yes/no/notbare" defaulting to "notbare"
> makes more sense.
This is on top of Junio's patch.
Documentation/config.txt | 13 +++++++------
Documentation/git-gc.txt | 4 ++--
git-gc.sh | 14 +++++++++-----
3 files changed, 18 insertions(+), 13 deletions(-)
diff --git a/Documentation/config.txt b/Documentation/config.txt
index 36b2c4f..3865535 100644
--- a/Documentation/config.txt
+++ b/Documentation/config.txt
@@ -322,13 +322,14 @@ format.headers::
by mail. See gitlink:git-format-patch[1].
gc.packrefs::
- `git gc` does not run `git pack-refs` by default
- so that older dumb-transport clients can still fetch
+ `git gc` does not run `git pack-refs` in a bare repository by
+ default so that older dumb-transport clients can still fetch
from the repository. Setting this to `true` lets `git
- gc` to run `git pack-refs`. Enable it only when you
- know you do not have to support such clients. The
- default will change to run `git pack-refs` in some
- future, and setting this to `false` will continue to
+ gc` to run `git pack-refs`. Setting this to `false` tells
+ `git gc` never to run `git pack-refs`. The default setting is
+ `notbare`. Enable it only when you know you do not have to
+ support such clients. The default setting will change to `true`
+ at some stage, and setting this to `false` will continue to
prevent `git pack-refs` from being run from `git gc`.
gc.reflogexpire::
diff --git a/Documentation/git-gc.txt b/Documentation/git-gc.txt
index ca38c2e..910f12a 100644
--- a/Documentation/git-gc.txt
+++ b/Documentation/git-gc.txt
@@ -64,8 +64,8 @@ kept. This defaults to 15 days.
The optional configuration variable 'gc.packrefs' determines if
`git gc` runs `git-pack-refs`. Without the configuration, `git-pack-refs`
-is not run by default to allow older dumb-transport clients
-fetch from the repository, but this will change in the future.
+is not run in bare repositories by default, to allow older dumb-transport
+clients fetch from the repository, but this will change in the future.
See Also
--------
diff --git a/git-gc.sh b/git-gc.sh
index 054f338..1a45de5 100755
--- a/git-gc.sh
+++ b/git-gc.sh
@@ -22,11 +22,15 @@ do
shift
done
-if pack_refs=$(git config --bool --get gc.packrefs) &&
- test "true" = "$pack_refs"
-then
- git-pack-refs --prune
-fi &&
+case "$(git config --get gc.packrefs)" in
+notbare|"")
+ test $(is_bare_repository) = true || pack_refs=true;;
+*)
+ pack_refs=$(git config --bool --get gc.packrefs)
+esac
+
+test "true" != "$pack_refs" ||
+git-pack-refs --prune &&
git-reflog expire --all &&
git-repack -a -d -l &&
$no_prune git-prune &&
--
1.5.0.rc4.2434.g472e7
^ permalink raw reply related [flat|nested] 9+ messages in thread
* Re: [PATCH] Do not run git-pack-refs by default for now from git-gc
2007-02-13 10:55 ` Johannes Schindelin
@ 2007-02-13 17:59 ` Junio C Hamano
0 siblings, 0 replies; 9+ messages in thread
From: Junio C Hamano @ 2007-02-13 17:59 UTC (permalink / raw)
To: Johannes Schindelin; +Cc: Jeff King, git
Thanks both.
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: linux-2.6.git/packed-refs???
2007-02-13 9:27 linux-2.6.git/packed-refs??? Junio C Hamano
2007-02-13 9:48 ` [PATCH] Do not run git-pack-refs by default for now from git-gc Junio C Hamano
@ 2007-02-13 18:16 ` Theodore Tso
2007-02-13 18:41 ` linux-2.6.git/packed-refs??? Johannes Schindelin
1 sibling, 1 reply; 9+ messages in thread
From: Theodore Tso @ 2007-02-13 18:16 UTC (permalink / raw)
To: Junio C Hamano; +Cc: torvalds, git
On Tue, Feb 13, 2007 at 01:27:03AM -0800, Junio C Hamano wrote:
> I noticed while scanning #git log that it appears that pack-refs
> was run in your public repo and some people cannot clone over
> dumb protocols with older git.
>
> commit 2986c02217f98809d8990e7679edf0f5d99f904d
> Author: Junio C Hamano <junkio@cox.net>
> Date: Wed Nov 22 22:24:09 2006 -0800
>
> git-fetch: fix dumb protocol transport to fetch from pack-pruned ref
>
> was the first revision that aligned dumb protocol clients with
> pack-ref; unfortunately anything older will not find the ref
> that was packed.
Stupid question. Suppose someone has run git pack-refs on a
repository. What is the recommended way to reverse the process. Is
the answer is to clone the repository and then throw away the original
one? Is there another way to do it?
- Ted
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: linux-2.6.git/packed-refs???
2007-02-13 18:16 ` linux-2.6.git/packed-refs??? Theodore Tso
@ 2007-02-13 18:41 ` Johannes Schindelin
0 siblings, 0 replies; 9+ messages in thread
From: Johannes Schindelin @ 2007-02-13 18:41 UTC (permalink / raw)
To: Theodore Tso; +Cc: Junio C Hamano, torvalds, git
Hi,
On Tue, 13 Feb 2007, Theodore Tso wrote:
> On Tue, Feb 13, 2007 at 01:27:03AM -0800, Junio C Hamano wrote:
> > I noticed while scanning #git log that it appears that pack-refs
> > was run in your public repo and some people cannot clone over
> > dumb protocols with older git.
> >
> > commit 2986c02217f98809d8990e7679edf0f5d99f904d
> > Author: Junio C Hamano <junkio@cox.net>
> > Date: Wed Nov 22 22:24:09 2006 -0800
> >
> > git-fetch: fix dumb protocol transport to fetch from pack-pruned ref
> >
> > was the first revision that aligned dumb protocol clients with
> > pack-ref; unfortunately anything older will not find the ref
> > that was packed.
>
> Stupid question. Suppose someone has run git pack-refs on a
> repository. What is the recommended way to reverse the process. Is
> the answer is to clone the repository and then throw away the original
> one? Is there another way to do it?
The easiest way I found here was to do
git-for-each-ref | while read sha1 tag name; do
git update-ref -m "undo pack-ref" $name $sha1
done
Beware: this is tested once, and copied from memory.
Ciao,
Dscho
^ permalink raw reply [flat|nested] 9+ messages in thread
end of thread, other threads:[~2007-02-13 18:41 UTC | newest]
Thread overview: 9+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2007-02-13 9:27 linux-2.6.git/packed-refs??? Junio C Hamano
2007-02-13 9:48 ` [PATCH] Do not run git-pack-refs by default for now from git-gc Junio C Hamano
2007-02-13 10:03 ` Johannes Schindelin
2007-02-13 10:29 ` Jeff King
2007-02-13 10:55 ` Johannes Schindelin
2007-02-13 17:59 ` Junio C Hamano
2007-02-13 13:01 ` [PATCH] git-gc: run pack-refs by default unless the repo is bare Johannes Schindelin
2007-02-13 18:16 ` linux-2.6.git/packed-refs??? Theodore Tso
2007-02-13 18:41 ` linux-2.6.git/packed-refs??? Johannes Schindelin
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.