On 06/21/2010 03:03 AM, Michael Nelson wrote:
> Additionally, afaik prior to SPRecipeBuilds process-upload was only
> called with sourcepackages with a cron invoked with:
> "/srv/launchpad.net/codelines/current/scripts/process-upload.py" as the
> lp_queue user.
Ah. It's misleading, then, that the behaviour is defined in a base class.
> I think it's a bug (or just wrong) that the file is being saved to the
> CWD just to be uploaded.
Agreed.
> If there's a quick permission fix to enable the
> SPRecipe builds to be uploaded - great (although I don't think it should
> be granting write access to the lp code root!)
I think that granting write access to the lp code root would be the only
possible permission fix.
We could update the way we invoke the script so that it runs in a temp
directory. That would be a hack, but it would mirror the way it works
on germanium, so I'd expect it to work.
>, but otherwise I think we
> should instead update unpackAndCheckSource() to check the changelog by
> moving it to a separate *tmp* directory, and upload it to the librarian
> from there.
Yes, I think this is the right solution.
Aaron
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On 06/21/2010 03:03 AM, Michael Nelson wrote: .net/codelines/ current/ scripts/ process- upload. py" as the
> Additionally, afaik prior to SPRecipeBuilds process-upload was only
> called with sourcepackages with a cron invoked with:
> "/srv/launchpad
> lp_queue user.
Ah. It's misleading, then, that the behaviour is defined in a base class.
> I think it's a bug (or just wrong) that the file is being saved to the
> CWD just to be uploaded.
Agreed.
> If there's a quick permission fix to enable the
> SPRecipe builds to be uploaded - great (although I don't think it should
> be granting write access to the lp code root!)
I think that granting write access to the lp code root would be the only
possible permission fix.
We could update the way we invoke the script so that it runs in a temp
directory. That would be a hack, but it would mirror the way it works
on germanium, so I'd expect it to work.
>, but otherwise I think we ource() to check the changelog by
> should instead update unpackAndCheckS
> moving it to a separate *tmp* directory, and upload it to the librarian
> from there.
Yes, I think this is the right solution.
Aaron enigmail. mozdev. org/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://
iEYEARECAAYFAkw fZ8YACgkQ0F+ nu1YWqI2SsACfb+ ggRV6aN0xJ5qI2H R1o1NVR hHQ48HPqKl9y13b k7
hIQAmwbXhas9WIv
=Bsjk
-----END PGP SIGNATURE-----