launchpad-dev team mailing list archive

[Aaron Bentley <aaron@xxxxxxxxxxxxx>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Curtis Hovey wrote:
> On Wed, 2009-11-25 at 12:45 -0500, Aaron Bentley wrote:
>> Would it make sense to have a policy of landing sampledata updates in
>> db-devel instead?  That's where I'd expect to find database changes, and
>> schema updates might have forced me to run "make schema" anyway.
> 
> Sampledata cannot be landed in devel, only db-devel.

AFAIK, the prohibition is against landing schema changes on devel, not
sampledata changes.  If our policy also forbids landing sampledata
changes, could you give me a link please?

Certainly, we don't have any technical safeguards to prevent this from
happening, because it did happen.  If we decide that this should be a
firm policy, we can add some.

>> Alternatively, automated detection of stale sampledata would have caught
>> the issue early.
> 
> The intent of the lint it to make sure anyone that lands a schema change
> or sampledata change is forced to update all data. Engineers are clearly
> circumnavigating the check.

I'm not sure what you mean.  Could you expand that?

> Any changes to the sampledata used by the
> test runner needs needs careful scrutiny; we made a conscious decision
> to not use it any more.

Agreed.  I believe this is was a case where the sampledata was actually
bogus, and wgrant updated it to be more accurate.

Aaron
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAksOmCsACgkQ0F+nu1YWqI0x3gCghOWlhqccLS/4afQxGpwGsAuZ
2wcAn0pswbI6FHnGyckGvQsjzWgd//g1
=hTii
-----END PGP SIGNATURE-----