Mailing List Archive

>>>>> On Wed, 14 Jul 2021, Marek Szuba wrote:

> On the plus side, nothing in here that requires changing to work with
> the new EAPI. On the minus side, we still got many EAPI-5 and 6
> consumers of this eclass in the tree so no chance of dropping support
> for these two at this time.

Shouldn't virtual/fortran go into BDEPEND in EAPIs 7 and 8?

On 2021-07-14 13:02, Ulrich Mueller wrote:

> Shouldn't virtual/fortran go into BDEPEND in EAPIs 7 and 8?

Good point! I've created https://bugs.gentoo.org/802153 so that we do
not lose track of this, that said it is beyond the scope of the issue at
hand (the eclass will not behave any differently here under EAPI 8 than
it does under EAPI 7) so I'll leave my current patch as it is.

--
Marecki

> On 14 Jul 2021, at 13:43, Marek Szuba <marecki@gentoo.org> wrote:
>
> On 2021-07-14 13:02, Ulrich Mueller wrote:
>
>> Shouldn't virtual/fortran go into BDEPEND in EAPIs 7 and 8?
>
> Good point! I've created https://bugs.gentoo.org/802153 so that we do not lose track of this, that said it is beyond the scope of the issue at hand (the eclass will not behave any differently here under EAPI 8 than it does under EAPI 7) so I'll leave my current patch as it is.
>

But to introduce a fix, isn't it a _lot_ easier to do it at the point of a new EAPI? What's the rush to add it if we're not ensuring we're using the new features correctly?

On 2021-07-16 22:50, Sam James wrote:

> But to introduce a fix, isn't it a _lot_ easier to do it at the point of a new EAPI?

In general, IMHO only if we intend to preserve the old (incorrect)
behaviour for older EAPIs - which in this particular case was not needed
because I cannot think of someone having come to rely on the fact EAPI-7
ebuilds inheriting fortran-2 could not be cross-compiled.

In this particular case, the patch I submitted for review earlier on
today explicitly mentions neither EAPI 7 nor EAPI 8 - so not really, no.
In fact, I would argue that in case of eclasses requiring more work to
adapt to a new EAPI trying to fix old bugs at the same time could
distract reviewers from the EAPI adaptation itself.

--
Marecki