Re: [gentoo-pms] [PATCH] EAPI must be at least a single char.

[Zac Medico]

On 09/30/2012 03:22 PM, Brian Harring wrote:
> On Sun, Sep 30, 2012 at 12:52:40PM -0700, Zac Medico wrote:
>> On 09/30/2012 12:44 PM, Brian Harring wrote:
>>> tries to write a PM, likely fucking that up.  If what you were saying
>>> was the actual intention behind it, that assignment would've just been
>>> along the lines of EAPI=("[^"]*"|'[^']*'|[^\t ]); aka "here's how you
>>> grab what looks like an EAPI assignment".
>>
>> I would have preferred a regex that just matches any assignment like
>> this, but didn't feel like bikeshedding it, since the one that's
>> currently in the spec works in practice.
>
> Counter point; portage doesn't actually enforce the rules of a valid
> EAPI name; correct me if I'm wrong obviously, but in checking the
> source, didn't see any such validation.

It doesn't need to check that EAPI names are valid. It only needs to
check that they are supported, so that's what it does.

> If the regex were as I suggested, that would be a non issue and we'd
> have *guranteed* EAPI name compliance- else it wouldn't match the
> invalid EAPI, and would stop looking at that line (falling back to
> EAPI=0).
>
> Basically, I'd like y'all to spell out the actual gains of having it
> loose like this, especailly in light of the fact the majority PM, via
> relying on that alone, doesn't do EAPI value enforcement.

It's enforced by the fact that all supported EAPIs happen to have valid
EAPI values.

> If we used the regex I suggested in the second email, this issue goes
> away, and we remove a potential landmine.
>
> Clarify to me why that landmine should be left in place.

The current spec works in practice, and that's all I really care about.
Still, I'd be in favor of changing the regex to match practically
anything that looks like an EAPI assignment, regardless of the validity
of the EAPI value.
-- 
Thanks,
Zac