[gentoo-dev] Why aren't GSoC projects affecting ::gentoo discussed on regular mls?

[gentoo-dev] Why aren't GSoC projects affecting ::gentoo discussed on regular mls?

[Michał Górny]

[-- Attachment #1: Type: text/plain, Size: 1790 bytes --]

(GSoC project mentors in CC)

Hi,

I would like to ask our this year's GSoC mentors a single question: why
weren't the GSoC proposals given proper discussion on our regular
mailing lists *before* they were accepted?

I can understand that most developers in Gentoo don't really care about
GSoC.  However, both projects we have this year [1] involve major
changes to ::gentoo that -- by policy -- require prior RFC.  In case
of the BLAS/LAPACK project there was a RFC *after* the project was
accepted, that was never fully answered.  In case of the MPI project,
I'm not aware of any public RFC or announcement.

I believe such decisions put all of us in a very bad position.  There is
a major work going on, almost secretly.  In the end, we will either be
forced to accept the result even if it doesn't meet our expectations, or
reject it and turn GSoC into some kind of grotesque situation.

The former is of course unacceptable from my point of view.  It would
mean that one or two developers are able to abuse paid programs such
as GSoC to unilaterally push their preferences into Gentoo.  We would be
forced to accept them unconditionally just because 'it's a done deal'.

The latter means the students has wasted their summer doing work that's
not going anywhere.  This is certainly demotivating and a bad PR for
Gentoo.  I suppose it also reduces our chance of getting into GSoC
again, if Google finds out that GSoC is spent on code going to trash.

So, again, why do single developers unilaterally decide on which
projects third party money is spent, and never bother discussing whether
those projects are really applicable beforehand?

[1] https://summerofcode.withgoogle.com/organizations/6416323580526592/

-- 
Best regards,
Michał Górny


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 618 bytes --]

[gentoo-dev] Re: Why aren't GSoC projects affecting ::gentoo discussed on regular mls?

[Benda Xu]

[-- Attachment #1: Type: text/plain, Size: 2973 bytes --]

Dear Michał,

Michał Górny <mgorny@gentoo.org> writes:

> I would like to ask our this year's GSoC mentors a single question:
> why weren't the GSoC proposals given proper discussion on our regular
> mailing lists *before* they were accepted?

> I can understand that most developers in Gentoo don't really care about
> GSoC.  However, both projects we have this year [1] involve major
> changes to ::gentoo that -- by policy -- require prior RFC.  In case
> of the BLAS/LAPACK project there was a RFC *after* the project was
> accepted, that was never fully answered.  In case of the MPI project,
> I'm not aware of any public RFC or announcement.

The proposal has been discussed in regular mailing lists.  

  https://archives.gentoo.org/gentoo-science/message/4d0186acdce6df538a2740e0f1146ae6

At the proposal stage it was not sent to gentoo-dev, because I thought
only science project was relevant to BLAS/LAPACK.  Later we find it to
be affecting more ebuilds, thus the RFC was sent to gentoo-dev.

  https://archives.gentoo.org/gentoo-dev/message/d917547f7a9e1226fca63632a1e02026

> I believe such decisions put all of us in a very bad position.  There is
> a major work going on, almost secretly.  In the end, we will either be
> forced to accept the result even if it doesn't meet our expectations, or
> reject it and turn GSoC into some kind of grotesque situation.

Michał, you were overreacting to the word "GSoC" since our original RFC
at gentoo-dev.  Please, just ignore GSoC when you are executing your
experise of QA.  Gentoo should be developed independently, regardless of
whether any development effort is supported by 3rd party.

> The former is of course unacceptable from my point of view.  It would
> mean that one or two developers are able to abuse paid programs such
> as GSoC to unilaterally push their preferences into Gentoo.  We would be
> forced to accept them unconditionally just because 'it's a done deal'.

See above.

> The latter means the students has wasted their summer doing work that's
> not going anywhere.  This is certainly demotivating and a bad PR for
> Gentoo.  I suppose it also reduces our chance of getting into GSoC
> again, if Google finds out that GSoC is spent on code going to trash.

That's why we are working together to find the best solution and reach a
consensus.

> So, again, why do single developers unilaterally decide on which
> projects third party money is spent, and never bother discussing whether
> those projects are really applicable beforehand?

I will leave this question to our GSoC manager.

Personally I don't regard the GSoC selection and decision process
interesting to all the Gentoo devs.  If you are interested in GSoC and
would like share your ideas to recruit student enthusiasts, you are more
than welcomed to join our team.

> [1]
> https://summerofcode.withgoogle.com/organizations/6416323580526592/

Cheers,
Benda

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 832 bytes --]

Re: [gentoo-dev] Re: Why aren't GSoC projects affecting ::gentoo discussed on regular mls?

[Marek Szuba]

On 2019-06-27 04:16, Benda Xu wrote:

> Michał, you were overreacting to the word "GSoC" since our original RFC
> at gentoo-dev.  Please, just ignore GSoC when you are executing your
> experise of QA.  Gentoo should be developed independently, regardless of
> whether any development effort is supported by 3rd party.
[...]
> Personally I don't regard the GSoC selection and decision process
> interesting to all the Gentoo devs.

In my opinion Michał has got a very good point regarding potential PR
consequences of us rejecting GSoC work. Of course it can be done, I've
seen my share of student projects of various sort getting binned
immediately after implementation - but more often than not it shows a
lack of of foresight, at best, on behalf of institutions which requested
manpower for such projects. Yes, it is good for you to have eventually
brought this to -dev - but IMHO it really is too late. In the future, I
would STRONGLY advise having a general discussion before having even a
single line of code written.

-- 
MS

[gentoo-dev] Re: Why aren't GSoC projects affecting ::gentoo discussed on regular mls?

[Benda Xu]

Hi Marek,

Marek Szuba <marecki@gentoo.org> writes:

> On 2019-06-27 04:16, Benda Xu wrote:
>
>> Michał, you were overreacting to the word "GSoC" since our original RFC
>> at gentoo-dev.  Please, just ignore GSoC when you are executing your
>> experise of QA.  Gentoo should be developed independently, regardless of
>> whether any development effort is supported by 3rd party.
> [...]
>> Personally I don't regard the GSoC selection and decision process
>> interesting to all the Gentoo devs.
>
> In my opinion Michał has got a very good point regarding potential PR
> consequences of us rejecting GSoC work. 

I agree with you.

> Of course it can be done, I've seen my share of student projects of
> various sort getting binned immediately after implementation - but
> more often than not it shows a lack of of foresight, at best, on
> behalf of institutions which requested manpower for such projects.

Agreed, too.  It is the mentor's duty to facilitate a win-win outcome to
our student, Gentoo and Google, by being more careful on visions and
planning.

> Yes, it is good for you to have eventually brought this to -dev - but
> IMHO it really is too late. In the future, I would STRONGLY advise
> having a general discussion before having even a single line of code
> written.

In retrospect, I was overconfident about my technical judgement.  I will
take your advice.  Thank you, Marek.

Yours,
Benda