[development] Core modules group

Derek Wright drupal at dwwright.net
Tue Sep 11 17:04:48 UTC 2007

On Sep 11, 2007, at 7:20 AM, Chris Johnson wrote:

> Um, do we even have to worry about the packaging script?

Yes, but not necessarily to catch this particular error...

> Couldn't we just check the correctness of the .install file at CVS  
> check-in time?

.info file you mean.  Yes, we certainly could.

> It's possible to provide feedback at that point, and we do know which
> repository the commit is being made to, as well.

Indeed.  And, this would be a better place to catch this kind of  
error (aside from the hordes of bogus .info files already in the  
repo).  By the time someone got far enough to make a tag and a  
release node for it, it's more of a pain to correct.  If our CVS repo  
won't even let them commit a .info file with bogus stuff, it'll be  
much better all the way around.

On Sep 11, 2007, at 7:40 AM, Doug Green wrote:
> Do we do any similar checks at CVS check-in.

Not similar in the sense of checking the validity of the *contents*  
of any files, but we certainly test validity at check-in time, for  
example, to enforce all the access control stuff for each project.

> webchick once talked to me about using coder at check-in.  If it's  
> possible to run php at this point, I think that some subset of the  
> coder checks could be implemented at this point.

Not just possible, already happening.  The only difference is you  
need a copy of the file someone's trying to commit, which isn't  
something the existing php code cares about.  That's certainly easy  
to do, it just means that bolting on .info file validation or other  
coder-related checks won't be utterly trivial, it'll require a tiny  
bit more work (to get a copy of the file to the script doing the tests).

And, as with all good ideas, we shall now continue this discussion in  
the issue queue:


-Derek (dww)

More information about the development mailing list