On 18/05/2011, at 6:31 PM, Ben Lippmeier wrote:

> On 18/05/2011, at 6:17 PM, Erik de Castro Lopo wrote:

>> In what litrle spare time I have I have been working on debugging
>> a problem that only seems to arise on linux-powerpc (bug #5111).
>> 
>> Since I'm time poor and CPU rich, I thought it might be useful to
>> try and use a bisect appraoch to try and find when the bug was
>> introduced. Unfortunately, if I do a git checkout of an old
>> revision (I tried with a commit dated Jan 4th 2011) and try and
>> build it it fails to build because the other source trees grabbed
>> via the sync-all script is not in sync with the main ghc tree.
>> 
>> Does anyone have a way of grabbing a snapshot of the source
>> tree from some arbitrary date?
> 
> I don't think that's possible. I've made tarballs of GHC states in the past, 
> saving them explicitly because rolling-back would be too hard. It would be 
> nice if it was possible though.
> 
> The patch history doesn't provide a working build at every point anyway. 
> People push *bundles* of patches, and the build needs to work after the 
> bundle, but it doesn't need to work after every patch. As we don't record the 
> bundles, the best you could really do is to examine the timestamps of the 
> commit messages in the mailing list, I think.

Though perhaps the buildbot, on completing a successful build, could record the 
hash of the most recent patch in each of the repos -- if it doesn't already.

Ben.


_______________________________________________
Cvs-ghc mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/cvs-ghc

Reply via email to