How can I figure out what commits relate to a given diff?

How can I figure out what commits relate to a given diff?

[Steven A. Falco]

I have a patch file, and I'd like to figure out which commits were used
to produce it.

For example, one of the diffs in the patch file begins with:

diff --git a/Makefile b/Makefile
index 9ceadaa..ad31bc6 100644

Is there a way to map a blob SHA1 to a commit?  In this example, I'd
like to map 9ceadaa and ad31bc6 to their commits.  It seems easy to go
the other way, seeing what is in a commit, but I've not been able to
find a method for going "backwards" from a blob to a commit.

    Thanks,
    Steve

Re: How can I figure out what commits relate to a given diff?

[Miklos Vajna]

[-- Attachment #1: Type: text/plain, Size: 608 bytes --]

On Thu, May 22, 2008 at 11:21:34AM -0400, "Steven A. Falco" <sfalco@harris.com> wrote:
> Is there a way to map a blob SHA1 to a commit?  In this example, I'd
> like to map 9ceadaa and ad31bc6 to their commits.  It seems easy to go
> the other way, seeing what is in a commit, but I've not been able to
> find a method for going "backwards" from a blob to a commit.

Something like:

for i in `git rev-list HEAD`
do
        git ls-tree -r $i|grep -q cb4e8ed && echo "$i contains cb4e8ed"
done

could do it for you.

I don't think there is a --contains for blobs, like git branch
--contains works for commits.

[-- Attachment #2: Type: application/pgp-signature, Size: 197 bytes --]

Re: How can I figure out what commits relate to a given diff?

[Jeff King]

On Thu, May 22, 2008 at 11:21:34AM -0400, Steven A. Falco wrote:

> For example, one of the diffs in the patch file begins with:
> 
> diff --git a/Makefile b/Makefile
> index 9ceadaa..ad31bc6 100644
> 
> Is there a way to map a blob SHA1 to a commit?  In this example, I'd
> like to map 9ceadaa and ad31bc6 to their commits.  It seems easy to go
> the other way, seeing what is in a commit, but I've not been able to
> find a method for going "backwards" from a blob to a commit.

Miklos suggested a way of looking up trees and commits which contain
blobs, but you can also look for that specific change of blobs:

  git log --pretty=format:'%h %s' --raw |
    perl -00 -ne 'print if /9ceadaa.*ad31bc6/'

-Peff

Re: How can I figure out what commits relate to a given diff?

[Steven A. Falco]

Miklos Vajna wrote:
> On Thu, May 22, 2008 at 11:21:34AM -0400, "Steven A. Falco" <sfalco@harris.com> wrote:
>> Is there a way to map a blob SHA1 to a commit?  In this example, I'd
>> like to map 9ceadaa and ad31bc6 to their commits.  It seems easy to go
>> the other way, seeing what is in a commit, but I've not been able to
>> find a method for going "backwards" from a blob to a commit.
> 
> Something like:
> 
> for i in `git rev-list HEAD`
> do
>         git ls-tree -r $i|grep -q cb4e8ed && echo "$i contains cb4e8ed"
> done
> 
> could do it for you.
> 
> I don't think there is a --contains for blobs, like git branch
> --contains works for commits.

Thanks!  It found the commit I needed.

	Steve

perforce import: git-p4 memory usage, perforce import: git-p4 memory usage

[Luke Diamand]

Hi!

I'm trying to import part of a perforce repo with git-p4.

However, git-p4 appears to try to read all the repo into memory (and
moreover seems to need about twice as much memory as repo).

Once it runs out of swap, it dies (unsurprisingly).

I think it's failing in readP4Files(), where it appears to read the
entire repository in one go with "p4 -G -x - print".

Can I just rework this function to do stuff one file at a time? Or is
that dumb?

Thanks
Luke Diamand

Re: perforce import: git-p4 memory usage

[Andreas Ericsson]

Luke Diamand wrote:
> Hi!
> 
> I'm trying to import part of a perforce repo with git-p4.
> 
> However, git-p4 appears to try to read all the repo into memory (and
> moreover seems to need about twice as much memory as repo).
> 
> Once it runs out of swap, it dies (unsurprisingly).
> 
> I think it's failing in readP4Files(), where it appears to read the
> entire repository in one go with "p4 -G -x - print".
> 
> Can I just rework this function to do stuff one file at a time? Or is
> that dumb?
> 

It's probably a lot slower, and since git works with changesets it
probably won't work all that good unless you somehow construct some
middle-stage which you can then feed to git-fastimport or some such.

-- 
Andreas Ericsson                   andreas.ericsson@op5.se
OP5 AB                             www.op5.se
Tel: +46 8-230225                  Fax: +46 8-230231

Re: perforce import: git-p4 memory usage

[Logan Kennelly]

On May 23, 2008, at 3:07 AM, Luke Diamand wrote:

> However, git-p4 appears to try to read all the repo into memory (and
> moreover seems to need about twice as much memory as repo).
>
> Once it runs out of swap, it dies (unsurprisingly).
>
> I think it's failing in readP4Files(), where it appears to read the
> entire repository in one go with "p4 -G -x - print".
>
> Can I just rework this function to do stuff one file at a time? Or is
> that dumb?


I don't know if they were ever accepted into the mainline, but see the
series of patches submitted by Tommy Thorn.

http://article.gmane.org/gmane.comp.version-control.git/72342

-- 
                                                         Logan Kennelly
       ,,,
      (. .)
--ooO-(_)-Ooo--