[gentoo-dev] python.eclass EAPI 4 support, this gets really annoying

[gentoo-dev] python.eclass EAPI 4 support, this gets really annoying

["Paweł Hajdan, Jr."]

[-- Attachment #1: Type: text/plain, Size: 456 bytes --]

OK, so what are the _blocking_ reasons for no EAPI 4 support in
python.eclass yet?

I understand you have some complicated patches in flight etc etc, but
are they _required_ for the eclass not to break with EAPI 4?

My point is that I'd like to use pkg_pretend in packages that use
python.eclass and I can't (for a long time). Unless there are really
important reasons like breakages I think we should really make
python.eclass support EAPI=4.


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 203 bytes --]

Re: [gentoo-dev] python.eclass EAPI 4 support, this gets really annoying

[Matt Turner]

On Sat, Oct 1, 2011 at 3:16 PM, "Paweł Hajdan, Jr."
<phajdan.jr@gentoo.org> wrote:
> OK, so what are the _blocking_ reasons for no EAPI 4 support in
> python.eclass yet?
>
> I understand you have some complicated patches in flight etc etc, but
> are they _required_ for the eclass not to break with EAPI 4?
>
> My point is that I'd like to use pkg_pretend in packages that use
> python.eclass and I can't (for a long time). Unless there are really
> important reasons like breakages I think we should really make
> python.eclass support EAPI=4.

Two weeks and no response from python@?

Re: [gentoo-dev] python.eclass EAPI 4 support, this gets really annoying

[Brian Harring]

On Fri, Oct 14, 2011 at 03:29:19PM -0400, Matt Turner wrote:
> On Sat, Oct 1, 2011 at 3:16 PM, "Pawe?? Hajdan, Jr."
> <phajdan.jr@gentoo.org> wrote:
> > OK, so what are the _blocking_ reasons for no EAPI 4 support in
> > python.eclass yet?
> >
> > I understand you have some complicated patches in flight etc etc, but
> > are they _required_ for the eclass not to break with EAPI 4?
> >
> > My point is that I'd like to use pkg_pretend in packages that use
> > python.eclass and I can't (for a long time). Unless there are really
> > important reasons like breakages I think we should really make
> > python.eclass support EAPI=4.
> 
> Two weeks and no response from python@?

You probably should've cc'd them.

You emailed gentoo-dev only.
~brian

Re: [gentoo-dev] python.eclass EAPI 4 support, this gets really annoying

["Paweł Hajdan, Jr."]

[-- Attachment #1: Type: text/plain, Size: 1016 bytes --]

On 10/14/11 12:39 PM, Brian Harring wrote:
> On Fri, Oct 14, 2011 at 03:29:19PM -0400, Matt Turner wrote:
>> On Sat, Oct 1, 2011 at 3:16 PM, "Pawe?? Hajdan, Jr."
>> <phajdan.jr@gentoo.org> wrote:
>>> OK, so what are the _blocking_ reasons for no EAPI 4 support in
>>> python.eclass yet?
>>>
>>> I understand you have some complicated patches in flight etc etc, but
>>> are they _required_ for the eclass not to break with EAPI 4?
>>>
>>> My point is that I'd like to use pkg_pretend in packages that use
>>> python.eclass and I can't (for a long time). Unless there are really
>>> important reasons like breakages I think we should really make
>>> python.eclass support EAPI=4.
>>
>> Two weeks and no response from python@?
> 
> You probably should've cc'd them.

CC-ing python@ then (but I expect the developers to follow gentoo-dev
anyway).

In case of no response, I plan to submit the thing to the council
agenda. I think the parts of python eclass that I use should work with
EAPI 4.


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 203 bytes --]

Re: [gentoo-dev] python.eclass EAPI 4 support, this gets really annoying

[Matthew Summers]

On Fri, Oct 14, 2011 at 4:20 PM, "Paweł Hajdan, Jr."
<phajdan.jr@gentoo.org> wrote:
> On 10/14/11 12:39 PM, Brian Harring wrote:
>> On Fri, Oct 14, 2011 at 03:29:19PM -0400, Matt Turner wrote:
>>> On Sat, Oct 1, 2011 at 3:16 PM, "Pawe?? Hajdan, Jr."
>>> <phajdan.jr@gentoo.org> wrote:
>>>> OK, so what are the _blocking_ reasons for no EAPI 4 support in
>>>> python.eclass yet?
>>>>
>>>> I understand you have some complicated patches in flight etc etc, but
>>>> are they _required_ for the eclass not to break with EAPI 4?
>>>>
>>>> My point is that I'd like to use pkg_pretend in packages that use
>>>> python.eclass and I can't (for a long time). Unless there are really
>>>> important reasons like breakages I think we should really make
>>>> python.eclass support EAPI=4.
>>>
>>> Two weeks and no response from python@?
>>
>> You probably should've cc'd them.
>
> CC-ing python@ then (but I expect the developers to follow gentoo-dev
> anyway).
>
> In case of no response, I plan to submit the thing to the council
> agenda. I think the parts of python eclass that I use should work with
> EAPI 4.
>
>

Its being worked on currently. There are many fairly difficult issues
to be worked through here. Your patience and/or contribution is
welcome.

-- 
Matthew W. Summers
Gentoo Foundation Inc.

Re: [gentoo-dev] python.eclass EAPI 4 support, this gets really annoying

[Matt Turner]

On Fri, Oct 14, 2011 at 5:25 PM, Matthew Summers
<quantumsummers@gentoo.org> wrote:
> Its being worked on currently. There are many fairly difficult issues
> to be worked through here.

That's kind of the question though. What's are the issues?

Re: [gentoo-dev] python.eclass EAPI 4 support, this gets really annoying

[Mike Gilbert]

[-- Attachment #1: Type: text/plain, Size: 1278 bytes --]

On 10/14/2011 5:28 PM, Matt Turner wrote:
> On Fri, Oct 14, 2011 at 5:25 PM, Matthew Summers
> <quantumsummers@gentoo.org> wrote:
>> Its being worked on currently. There are many fairly difficult issues
>> to be worked through here.
> 
> That's kind of the question though. What's are the issues?

The only issue that has been raised is that the eclass will die if
python_pkg_setup is not called from the ebuild. This can happen either
implicitly (the function is exported) or explicitly (by calling it from
pkg_setup).

I haven't gotten a very good explanation as to why this is necessary
from Arfrever. He has given some very short, non-informative
explanations like "Jython requires it".

Puzzling it out myself takes quite a bit of brain power, given the
complexity of the eclass. There may in fact be no reason for it at all.
Either way, somebody needs to actually understand it.

Other than that issue, I think we really just need to do some testing.
Despite there being a large number of people in the python herd, there
aren't many who actually have the time to do this.

It would be helpful if folks could brain storm a list of python related
packages to test under EAPI 4. I may have some time over the next few
weekends to work on this.


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 228 bytes --]

Re: [gentoo-dev] python.eclass EAPI 4 support, this gets really annoying

[Robin H. Johnson]

On Fri, Oct 14, 2011 at 05:52:59PM -0400, Mike Gilbert wrote:
> It would be helpful if folks could brain storm a list of python related
> packages to test under EAPI 4. I may have some time over the next few
> weekends to work on this.
dev-vcs/git needs python eclasses to have EAPI4 so REQUIRED_USE can be
used to solve bug #353657.

-- 
Robin Hugh Johnson
Gentoo Linux: Developer, Trustee & Infrastructure Lead
E-Mail     : robbat2@gentoo.org
GnuPG FP   : 11AC BA4F 4778 E3F6 E4ED  F38E B27B 944E 3488 4E85

Re: [gentoo-dev] python.eclass EAPI 4 support, this gets really annoying

[Tony "Chainsaw" Vroon]

[-- Attachment #1: Type: text/plain, Size: 231 bytes --]

On Fri, 2011-10-14 at 22:53 +0000, Robin H. Johnson wrote:
> dev-vcs/git needs python eclasses to have EAPI4 so REQUIRED_USE can be
> used to solve bug #353657. 

Similar problems in dev-vcs/subversion...

Regards,
Tony V.

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Re: [gentoo-dev] python.eclass EAPI 4 support, this gets really annoying

[Thomas Sachau]

[-- Attachment #1: Type: text/plain, Size: 1395 bytes --]

"Paweł Hajdan, Jr." schrieb:
> On 10/14/11 12:39 PM, Brian Harring wrote:
>> On Fri, Oct 14, 2011 at 03:29:19PM -0400, Matt Turner wrote:
>>> On Sat, Oct 1, 2011 at 3:16 PM, "Pawe?? Hajdan, Jr."
>>> <phajdan.jr@gentoo.org> wrote:
>>>> OK, so what are the _blocking_ reasons for no EAPI 4 support in
>>>> python.eclass yet?
>>>>
>>>> I understand you have some complicated patches in flight etc etc, but
>>>> are they _required_ for the eclass not to break with EAPI 4?
>>>>
>>>> My point is that I'd like to use pkg_pretend in packages that use
>>>> python.eclass and I can't (for a long time). Unless there are really
>>>> important reasons like breakages I think we should really make
>>>> python.eclass support EAPI=4.
>>>
>>> Two weeks and no response from python@?
>>
>> You probably should've cc'd them.
> 
> CC-ing python@ then (but I expect the developers to follow gentoo-dev
> anyway).
> 
> In case of no response, I plan to submit the thing to the council
> agenda. I think the parts of python eclass that I use should work with
> EAPI 4.
> 

What do you expect the council to do?

Neither you nor the council can force anyone to do anything.

The only thing you can do is to kindly ask about remaining issues and helping with solving them. And
i am sure, that everyone would like to see some help to understand and improve the python eclass. :-)


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 380 bytes --]

Re: [gentoo-dev] python.eclass EAPI 4 support, this gets really annoying

["Paweł Hajdan, Jr."]

[-- Attachment #1: Type: text/plain, Size: 692 bytes --]

On 10/14/11 3:32 PM, Thomas Sachau wrote:
> What do you expect the council to do?

Say it's OK to make python.eclass not die on EAPI-4. At least my use
case will not become broken by this.

> Neither you nor the council can force anyone to do anything.

Not really. It should always be possible to escalate painful issues like
this.

> The only thing you can do is to kindly ask about remaining issues and helping with solving them. And
> i am sure, that everyone would like to see some help to understand and improve the python eclass. :-)

Maybe we should start over. Seriously, I'm considering proposing some
lightweight eclass for python-dependent packages that *works*


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 203 bytes --]

Re: [gentoo-dev] python.eclass EAPI 4 support, this gets really annoying

[Alec Warner]

On Fri, Oct 14, 2011 at 3:51 PM, "Paweł Hajdan, Jr."
<phajdan.jr@gentoo.org> wrote:
> On 10/14/11 3:32 PM, Thomas Sachau wrote:
>> What do you expect the council to do?
>
> Say it's OK to make python.eclass not die on EAPI-4. At least my use
> case will not become broken by this.
>
>> Neither you nor the council can force anyone to do anything.
>
> Not really. It should always be possible to escalate painful issues like
> this.

I believe op's point is that there is no one to escalate the problem
to; certainly the council members are not going to do the work
themselves and we already have our best people on it.

-A

>
>> The only thing you can do is to kindly ask about remaining issues and helping with solving them. And
>> i am sure, that everyone would like to see some help to understand and improve the python eclass. :-)
>
> Maybe we should start over. Seriously, I'm considering proposing some
> lightweight eclass for python-dependent packages that *works*
>
>

Re: [gentoo-dev] python.eclass EAPI 4 support, this gets really annoying

["Paweł Hajdan, Jr."]

[-- Attachment #1: Type: text/plain, Size: 967 bytes --]

On 10/14/11 5:38 PM, Alec Warner wrote:
> I believe op's point is that there is no one to escalate the problem
> to; certainly the council members are not going to do the work
> themselves and we already have our best people on it.

I'm aware of that. My point is that I think there are many scenarios in
which EAPI-4 + python.eclass can work, especially if it's used only for
few things in cases like www-client/chromium

Because the python team takes _ages_ to do the transition that is
holding back many other packages, because they've made python.eclass
overly complex and now try to make it perfect,

I'd just like to get an "OK" to enable EAPI-4 for that eclass.

Please note that it's still up to dependent packages which EAPI they
use. If they break python.eclass with EAPI-4 they shouldn't update to
that EAPI. However, if there are packages using python.eclass that could
work fine with EAPI-4, it shouldn't be blocking them for *ages*


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 203 bytes --]

Re: [gentoo-dev] python.eclass EAPI 4 support, this gets really annoying

[Mike Gilbert]

[-- Attachment #1: Type: text/plain, Size: 1354 bytes --]

On 10/14/2011 09:11 PM, "Paweł Hajdan, Jr." wrote:
> On 10/14/11 5:38 PM, Alec Warner wrote:
>> I believe op's point is that there is no one to escalate the problem
>> to; certainly the council members are not going to do the work
>> themselves and we already have our best people on it.
> 
> I'm aware of that. My point is that I think there are many scenarios in
> which EAPI-4 + python.eclass can work, especially if it's used only for
> few things in cases like www-client/chromium
> 
> Because the python team takes _ages_ to do the transition that is
> holding back many other packages, because they've made python.eclass
> overly complex and now try to make it perfect,
> 
> I'd just like to get an "OK" to enable EAPI-4 for that eclass.
> 
> Please note that it's still up to dependent packages which EAPI they
> use. If they break python.eclass with EAPI-4 they shouldn't update to
> that EAPI. However, if there are packages using python.eclass that could
> work fine with EAPI-4, it shouldn't be blocking them for *ages*
> 

That would be an ok approach from my perspective: We could just change
line 14 of python.eclass and let package maintainers report breakage as
they increment EAPI. I am confident that nothing EAPI <= 3 would break.

Is anyone (especially djc and the python herd members) opposed to this?


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 230 bytes --]

Re: [gentoo-dev] python.eclass EAPI 4 support, this gets really annoying

[Dirkjan Ochtman]

On Sat, Oct 15, 2011 at 03:54, Mike Gilbert <floppym@gentoo.org> wrote:
> That would be an ok approach from my perspective: We could just change
> line 14 of python.eclass and let package maintainers report breakage as
> they increment EAPI. I am confident that nothing EAPI <= 3 would break.
>
> Is anyone (especially djc and the python herd members) opposed to this?

Seems fine to me; I can't really think of a practical better way.

Cheers,

Dirkjan

Re: [gentoo-dev] python.eclass EAPI 4 support, this gets really annoying

["Paweł Hajdan, Jr."]

[-- Attachment #1: Type: text/plain, Size: 763 bytes --]

On 10/15/11 2:42 AM, Dirkjan Ochtman wrote:
> On Sat, Oct 15, 2011 at 03:54, Mike Gilbert <floppym@gentoo.org> wrote:
>> That would be an ok approach from my perspective: We could just change
>> line 14 of python.eclass and let package maintainers report breakage as
>> they increment EAPI. I am confident that nothing EAPI <= 3 would break.
>>
>> Is anyone (especially djc and the python herd members) opposed to this?
> 
> Seems fine to me; I can't really think of a practical better way.

Thank you, change committed to CVS then. Hopefully nobody will get upset
about this.

I'll wait a few days before I start using EAPI-4 in ebuilds using
python.eclass, but I've done local tests and everything works fine (for
the ebuild I (co-)maintain).


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 203 bytes --]

Re: [gentoo-dev] python.eclass EAPI 4 support, this gets really annoying

[Samuli Suominen]

On 10/16/2011 12:00 AM, "Paweł Hajdan, Jr." wrote:
> On 10/15/11 2:42 AM, Dirkjan Ochtman wrote:
>> On Sat, Oct 15, 2011 at 03:54, Mike Gilbert <floppym@gentoo.org> wrote:
>>> That would be an ok approach from my perspective: We could just change
>>> line 14 of python.eclass and let package maintainers report breakage as
>>> they increment EAPI. I am confident that nothing EAPI <= 3 would break.
>>>
>>> Is anyone (especially djc and the python herd members) opposed to this?
>>
>> Seems fine to me; I can't really think of a practical better way.
> 
> Thank you, change committed to CVS then. Hopefully nobody will get upset
> about this.
> 
> I'll wait a few days before I start using EAPI-4 in ebuilds using
> python.eclass, but I've done local tests and everything works fine (for
> the ebuild I (co-)maintain).
> 

Thanks, this is most appericiated.   This allowed me to kill EAPI=3
support from xfconf.eclass.

Re: [gentoo-dev] python.eclass EAPI 4 support, this gets really annoying

[Jesus Rivero (Neurogeek)]

On Fri, Oct 14, 2011 at 9:54 PM, Mike Gilbert <floppym@gentoo.org> wrote:
> On 10/14/2011 09:11 PM, "Paweł Hajdan, Jr." wrote:
>> On 10/14/11 5:38 PM, Alec Warner wrote:
>>> I believe op's point is that there is no one to escalate the problem
>>> to; certainly the council members are not going to do the work
>>> themselves and we already have our best people on it.
>>
>> I'm aware of that. My point is that I think there are many scenarios in
>> which EAPI-4 + python.eclass can work, especially if it's used only for
>> few things in cases like www-client/chromium
>>
>> Because the python team takes _ages_ to do the transition that is
>> holding back many other packages, because they've made python.eclass
>> overly complex and now try to make it perfect,
>>
>> I'd just like to get an "OK" to enable EAPI-4 for that eclass.
>>
>> Please note that it's still up to dependent packages which EAPI they
>> use. If they break python.eclass with EAPI-4 they shouldn't update to
>> that EAPI. However, if there are packages using python.eclass that could
>> work fine with EAPI-4, it shouldn't be blocking them for *ages*
>>
>
> That would be an ok approach from my perspective: We could just change
> line 14 of python.eclass and let package maintainers report breakage as
> they increment EAPI. I am confident that nothing EAPI <= 3 would break.
>
> Is anyone (especially djc and the python herd members) opposed to this?
>
>

Sorry I wasn't able to post before. But...
This can be done and in fact has been discussed before, just allow
ebuild to not die with EAPI=4, but this doesn do anything at all, just
not die on EAPI=4. All the features and the good stuff just won't be
there as other use cases need (as Robin and Tony mentioned).

We've been working on a redesign of the eclass but is nothing like
stealing candy from a kid, there are many things involved, such as the
large amount of Python ABIs that people use regularly, distutils
quirks, current eclass complexity, among others that make it quite
challenging to come up with something new.

I'm all up for making eclass accept EAPI=4 ebuilds, but to fully
implement EAPI=4 fesatures, I'm going to have to ask you guys for a
bit of more patience. I know you have done just that, being patient,
but just a bit more. I know we can deliver a solution for this soon
enough.

Best regards,

-- 
Jesus Rivero (Neurogeek)
Gentoo Developer