launchpad-dev team mailing list archive

[Henning Eggers <henning.eggers@xxxxxxxxxxxxx>]

Am 20.08.2010 10:44, schrieb Stuart Bishop:
> On Fri, Aug 20, 2010 at 3:19 AM, Francis J. Lacoste
>> Running make clean and trying to build locally should reproduce the failure.
> 
> Oh dear. I suspect there might be more work involved than that. In
> some cases, imports are in an explicit order to avoid circular import
> issues.

I am still holding my breath and see what happens. I fear that this might
still hit me. I already included code in the script to skip a file completely
(done for c/l/interfaces/__init__.py) or to explicitly place imports at the top.

The problem here was that the mailman build failure was masking the real
errors when building locally. I got around that by *not* doing a "make clean"
after such a failure but just another "make" which would then assume mailman
was build and proceed - showing me the real error.

The real errors, btw, were the script being too optimistic about the current
code adhering to conventions (e.g. 4 space indentions) and the aforementioned
ordering of imports. It may well be that this remains the only problem.

Good news: The tree is building locally now, so I sent it off to ec2 for some
testing and even possibly submitting to PQM, if it passes. I don't get my
hopes up about that yet, though.

I am on devel r11403.

Cheers,
Henning