On Tue, Apr 14, 2009 at 4:03 AM, Carl-Daniel Hailfinger < c-d.hailfinger.devel.2006@gmx.net> wrote:
On 14.04.2009 03:29, Richard Smith wrote:
Peter Stuge wrote:
ron minnich wrote:
did you test with abuild :-)
No sir. I have neither procedure nor CPU power for abuild. :\
By the time I would have managed to run abuild once, the server would probably have run abuild thrice, including the two potential rounds of fixes.
Something I've always wondered about is why is abuild only in response to a svn commit? I think it would be quite handy if you had something like abuild-v[23]@coreboot.org and any patch you send as an attachment to it pulls a copy of the tree applies the patch and then runs abuild on the tree and then emails back the results.
Security reasons? What's stopping anyone from mailing a patch which starts a shell on the abuild server which listens on port 12345 or similar fun?
I'm not understanding how that could happen, unless they sent a patch against abuild itself, and that could be easily rectified by running patch as another user, and having abuild owned by e.g. root and marked as rx for others. That would kill the ability to have patches against abuild tested by abuild, but those are so rare I don't think it'd be a big issue. I'm considering setting up something like this, my Q6600 with 6GB of ram runs idle most of the day, I'm just dreading setting up a mail system on it to send out the results.
-Corey