proposal: add "packagers" field (list of strings (names)) to package definition

proposal: add "packagers" field (list of strings (names)) to package definition

[Florian Paul Schmidt]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256


...and encourage its use. The intended semantics is to list people
that have contributed to the packaging effort. The motivation behind
this proposal is that in many free software projects attribution can
be a major source of motivation to get people involved. Having the
packagers be first class citizens in the package definitions (as
opposed to the information being only implicitly available through
e.g. "git blame") would allow things like "guix package" or the
package list on the website to display the contributor's names.

And if in a standard format containing additional info like an email
address then bug reports for a package might even get CC'ed
automatically to the contributors (though this might have some privacy
implications - but providing an email address or even any entry in the
packagers field is purely opt-in).

Note that the (largely unused) maintainer field [1] plays a slightly
different role - judging purely from its name, since it's not too well
documented: A maintainer has responsibilities beyond the role of a
contributor.

[1] https://www.gnu.org/software/guix/manual/guix.html#package-Reference

WDYT?

Regards,
Flo


- -- 
https://fps.io
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2

iQEcBAEBCAAGBQJWXVZbAAoJEA5f4Coltk8ZwNMH/2HLaUhO+j4U3dYLR4BhKvcN
YKLd7lJyIRXYHCgjePny1avV+QwUrUEyDz22EE4Ucktfm85wHxoILWN4eIPiCLno
SQCEiqkVExRGzqbVzTfpqtEbhpxv8xT/vnR+IGNjS7PBwAODVar2jv3RIrPLhS0+
2gLeMUemHg21isfJh+eREYybYdYX6KKDTOfJFAoKZh3y7HB1QuO7JtCj7rUx+pKz
X2vRovQYJobg+9IMbmDum4v7+ptXV6fKc0P0z4aB1QGmGyiyok9nFDifNEw7CiE/
9XaV30fwYPGeZqJ1SPWm9Pc80OXM+gEOrVWemW/fgFOYiIz3JkeEjz6aKIS9km8=
=FmOO
-----END PGP SIGNATURE-----

Re: proposal: add "packagers" field (list of strings (names)) to package definition

[Florian Paul Schmidt]

On 12/01/2015 09:12 AM, Florian Paul Schmidt wrote:
> 
> ...and encourage its use. The intended semantics is to list people 
> that have contributed to the packaging effort. The motivation
> behind

[...]

Something like:

diff --git a/guix/packages.scm b/guix/packages.scm
index 68fb091..efe1dbf 100644
--- a/guix/packages.scm
+++ b/guix/packages.scm
@@ -78,6 +78,7 @@
             package-home-page
             package-supported-systems
             package-maintainers
+            package-packagers
             package-properties
             package-location
             package-field-location
@@ -266,6 +267,9 @@ name of its URI."
                      (default %supported-systems))
   (maintainers package-maintainers (default '()))

+  (packagers package-packagers (default '()))     ; list of people
that worked
+                                                  ; on the package
+
   (properties package-properties (default '()))   ; alist for
anything else

   (location package-location


Regards,
Flo

-- 
https://fps.io

Re: proposal: add "packagers" field (list of strings (names)) to package definition

[Leo Famulari]

On Tue, Dec 01, 2015 at 09:12:12AM +0100, Florian Paul Schmidt wrote:
> ...and encourage its use. The intended semantics is to list people
> that have contributed to the packaging effort. The motivation behind
> this proposal is that in many free software projects attribution can
> be a major source of motivation to get people involved. Having the
> packagers be first class citizens in the package definitions (as
> opposed to the information being only implicitly available through
> e.g. "git blame") would allow things like "guix package" or the
> package list on the website to display the contributor's names.

All the contributors do get attribution in the copyright notice at the
top of each file, although that information is not linked to their
actual contributions except through git.

> And if in a standard format containing additional info like an email
> address then bug reports for a package might even get CC'ed
> automatically to the contributors (though this might have some privacy
> implications - but providing an email address or even any entry in the
> packagers field is purely opt-in).

I like the idea of using this information programatically.

> WDYT?

The nice thing about `git blame` is that it's "never wrong" — you can
easily find out who is actually invested in the relevant code based on
their actions, rather than what they claimed when putting their name in
the "maintainer" or "packager" field. That is, `git blame` shows
revealed preferences while the "maintainer" field shows rhetorical
preferences. Maybe `git blame` gets stale, but you can judge freshness
based on the age of the commits.

Plus I can see some "political" issues in the future where people lay
claim to parts of the code base and justify it based on their name being in
the packager field. Personally, I think we should avoid creating these
sorts of bureaucracies if its not necessary.

I noticed that the NixOS github has a "mention-bot" that
automatically contacts people based on `git blame` if their old code is
subject to a pull request. You can see it in action here:
https://github.com/NixOS/nixpkgs/pull/11329

I think we should let the git repository be the single source of truth
for figuring out who is responsible for the code. If necessary, we can
build some automation around the git repo.

Thoughts?

Re: proposal: add "packagers" field (list of strings (names)) to package definition

[Leo Famulari]

I should add a disclaimer: take my point of view with a grain of salt. There are other Guix contributors with a much broader perspective on the project. I'm still a newbie!

On December 1, 2015 12:35:18 PM EST, Leo Famulari <leo@famulari.name> wrote:
>On Tue, Dec 01, 2015 at 09:12:12AM +0100, Florian Paul Schmidt wrote:
>> ...and encourage its use. The intended semantics is to list people
>> that have contributed to the packaging effort. The motivation behind
>> this proposal is that in many free software projects attribution can
>> be a major source of motivation to get people involved. Having the
>> packagers be first class citizens in the package definitions (as
>> opposed to the information being only implicitly available through
>> e.g. "git blame") would allow things like "guix package" or the
>> package list on the website to display the contributor's names.
>
>All the contributors do get attribution in the copyright notice at the
>top of each file, although that information is not linked to their
>actual contributions except through git.
>
>> And if in a standard format containing additional info like an email
>> address then bug reports for a package might even get CC'ed
>> automatically to the contributors (though this might have some
>privacy
>> implications - but providing an email address or even any entry in
>the
>> packagers field is purely opt-in).
>
>I like the idea of using this information programatically.
>
>> WDYT?
>
>The nice thing about `git blame` is that it's "never wrong" — you can
>easily find out who is actually invested in the relevant code based on
>their actions, rather than what they claimed when putting their name in
>the "maintainer" or "packager" field. That is, `git blame` shows
>revealed preferences while the "maintainer" field shows rhetorical
>preferences. Maybe `git blame` gets stale, but you can judge freshness
>based on the age of the commits.
>
>Plus I can see some "political" issues in the future where people lay
>claim to parts of the code base and justify it based on their name
>being in
>the packager field. Personally, I think we should avoid creating these
>sorts of bureaucracies if its not necessary.
>
>I noticed that the NixOS github has a "mention-bot" that
>automatically contacts people based on `git blame` if their old code is
>subject to a pull request. You can see it in action here:
>https://github.com/NixOS/nixpkgs/pull/11329
>
>I think we should let the git repository be the single source of truth
>for figuring out who is responsible for the code. If necessary, we can
>build some automation around the git repo.
>
>Thoughts?

Re: proposal: add "packagers" field (list of strings (names)) to package definition

[Florian Paul Schmidt]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

On 12/01/2015 06:35 PM, Leo Famulari wrote:

> All the contributors do get attribution in the copyright notice at
> the top of each file, although that information is not linked to
> their actual contributions except through git.

If they feel their contribution is significant enough to even warrant
an entry there.

[snip]

> Plus I can see some "political" issues in the future where people
> lay claim to parts of the code base and justify it based on their
> name being in the packager field. Personally, I think we should
> avoid creating these sorts of bureaucracies if its not necessary.

Sure, there's a tradeoff there. If the intended semantics are
communicated clearly I don't see that problem.

> I noticed that the NixOS github has a "mention-bot" that 
> automatically contacts people based on `git blame` if their old
> code is subject to a pull request. You can see it in action here: 
> https://github.com/NixOS/nixpkgs/pull/11329
> 
> I think we should let the git repository be the single source of
> truth for figuring out who is responsible for the code. If
> necessary, we can build some automation around the git repo.

Yeah, I like the nixos workflow quite a bit. github is a pretty nice
tool, but sadly nonfree, thus a machina non grata for guix. The nixos
devs actively encourage people to add themselves to the maintainer
fields btw..

Flo



- -- 
https://fps.io
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2

iQEcBAEBCAAGBQJWX22gAAoJEA5f4Coltk8Zb+8H/3qprqfKFCzK6049IGMm5EZl
+RVZICpFcvZQH8QLsNeBrkiDfqEq/lTDg7A6emW38GttxSTE4OCy8rUNy9y73TDB
WdSaglazbYHH3HKv4vIZQSTmkODDkuHS0HtTjgTxqicOSwZBChyhEP2jW04rvMCg
ihSigW/MvASkLSwUH2ridL5kHyfDCzdipOy9WnysLsAlTVJi7SUwLvk6hAx2IP3l
km7qm0ZDFSwglM5NeUHitYHr8eZ0GJLTXNi+lhucXucg+OgpR3pozTkTpX5XjEUG
TnhxnlNhLCt2tGL9LWpRVqE80va8SHON96q8HX3XYy2RxIzpG2mds7ue5RNQcJM=
=IUiM
-----END PGP SIGNATURE-----

Re: proposal: add "packagers" field (list of strings (names)) to package definition

[Andy Wingo]

On Tue 01 Dec 2015 08:12, Florian Paul Schmidt <mista.tapas@gmx.net> writes:

> ...and encourage its use. The intended semantics is to list people
> that have contributed to the packaging effort. The motivation behind
> this proposal is that in many free software projects attribution can
> be a major source of motivation to get people involved. Having the
> packagers be first class citizens in the package definitions (as
> opposed to the information being only implicitly available through
> e.g. "git blame") would allow things like "guix package" or the
> package list on the website to display the contributor's names.

Since you ask for opinions, -1 from me :)

Currently Guix packages are more-or-less collectively owned.
Introducing this field implies to me an introduction of ownership of
packages.  Ownership has a number of negative effects: it can inhibit
casual fixes and it can introduce unnecessary conflicts.

The kind of ego-based positive feedback that having a maintainer field
would introduce is negative IMO.  It is good if people feel attached to
Guix-the-project and GuixSD-the-software-distribution, and to a degree
they get that by copyright, git blame, and NEWS.  Feeling attached to a
particular package is not as useful for the project or the distribution.

IMO :)

Andy

Re: proposal: add "packagers" field (list of strings (names)) to package definition

[Thompson, David]

On Wed, Dec 2, 2015 at 8:43 AM, Andy Wingo <wingo@igalia.com> wrote:
> On Tue 01 Dec 2015 08:12, Florian Paul Schmidt <mista.tapas@gmx.net> writes:
>
>> ...and encourage its use. The intended semantics is to list people
>> that have contributed to the packaging effort. The motivation behind
>> this proposal is that in many free software projects attribution can
>> be a major source of motivation to get people involved. Having the
>> packagers be first class citizens in the package definitions (as
>> opposed to the information being only implicitly available through
>> e.g. "git blame") would allow things like "guix package" or the
>> package list on the website to display the contributor's names.
>
> Since you ask for opinions, -1 from me :)
>
> Currently Guix packages are more-or-less collectively owned.
> Introducing this field implies to me an introduction of ownership of
> packages.  Ownership has a number of negative effects: it can inhibit
> casual fixes and it can introduce unnecessary conflicts.
>
> The kind of ego-based positive feedback that having a maintainer field
> would introduce is negative IMO.  It is good if people feel attached to
> Guix-the-project and GuixSD-the-software-distribution, and to a degree
> they get that by copyright, git blame, and NEWS.  Feeling attached to a
> particular package is not as useful for the project or the distribution.

+1

- Dave

Re: proposal: add "packagers" field (list of strings (names)) to package definition

[Florian Paul Schmidt]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

On 12/02/2015 03:09 PM, Thompson, David wrote:

>> The kind of ego-based positive feedback that having a maintainer
>> field would introduce is negative IMO.  It is good if people feel
>> attached to Guix-the-project and
>> GuixSD-the-software-distribution, and to a degree they get that
>> by copyright, git blame, and NEWS.  Feeling attached to a 
>> particular package is not as useful for the project or the
>> distribution.
> 
> +1

Aww, too bad. I think it's a missed opportunity to be inclusive and
generous with attribution which might motivate people to contribute.
Guix is in a unique position with this opportunity.

Flo

- -- 
https://fps.io
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2

iQEcBAEBCAAGBQJWXwdnAAoJEA5f4Coltk8Zx/EH/i+uevBW3FBLrBShGUYPjn6k
RqkTcAqneBc26PR8KxNboTSwxEbrtaqpSwHREtupZvRB5lV4P+2L7WKQQQe1ZlWx
oG8b7g8yO9SMprqz3sqKkLmofFlXZuodygji4acAWvbHfbqeJAhfUvqsDEHf4Ilj
rNjVZAigB8TXyIIhwRIPqUfVjgPtnT01wpa8igTma1WOfgXg9nhpAO9FeeoUmAYN
QvrnNlmQs+/DGKemSp8uzgznlRlJ4OOvHH6tFMwZMR7DS6xI0kWvW/HDmEvXswHX
mJiyZAT3GFS+2B95BL+lIP4PXwpi3gP8Y35VwvpO7qSdWgCFX/UaUgNpw8n1o7U=
=M4qg
-----END PGP SIGNATURE-----

Re: proposal: add "packagers" field (list of strings (names)) to package definition

[Andy Wingo]

On Wed 02 Dec 2015 14:59, Florian Paul Schmidt <mista.tapas@gmx.net> writes:

> Aww, too bad. I think it's a missed opportunity to be inclusive and
> generous with attribution which might motivate people to contribute.
> Guix is in a unique position with this opportunity.

If there are other ways we can recognize people's work, we should
consider them :)  Think of it this way though, a packager is a person
who packages.  The value is in the verb, not the noun; packaging is an
ongoing process.  We should reward the process, via NEWS, and perhaps in
other ways.

Besides, no one wants to be the person that removes a name from a list
of a package's packagers!

Andy

Re: proposal: add "packagers" field (list of strings (names)) to package definition

[Florian Paul Schmidt]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

On 12/02/2015 05:12 PM, Andy Wingo wrote:

> If there are other ways we can recognize people's work, we should 
> consider them :)  Think of it this way though, a packager is a
> person who packages.  The value is in the verb, not the noun;
> packaging is an ongoing process.  We should reward the process, via
> NEWS, and perhaps in other ways.
> 
> Besides, no one wants to be the person that removes a name from a
> list of a package's packagers!

Why remove a name that made it into the list ever? Did the package
packagers package packages somewhere in the past or didn't they (or
contributed to packaging packages - contest: What sentence can include
the most occurences of "package" and still make sense)?

Like I said in the other reply: It's just a proposal to ponder a bit.
Maybe it does make sense, maybe it doesn't. To me it still seems it does
..

Flo

- -- 
https://fps.io
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2

iQEcBAEBCAAGBQJWXzvLAAoJEA5f4Coltk8ZHTwIALSvOUmeZysjHbu9TDM8z7LK
KFMJIlf0fe7bmVn/O8ynTiKbPHS3MF9xmNBADPdilQ/dIs7AgZkvaoQxoIogTFjP
ImM1HFiaTV1RagRv+0f+zKD3s16+duDtBKEdu5wcIUVtOl1pmtaO8sNeXLvRnjFB
ZTUxqy/s4/OhsH3XA/IbbXBUc/9uzgunNynQTAyakXvH3VhOdj1jFQxaMU4EfcwI
aF3Q9TVJDAw3kGH6r8eKXLte90d2GZpScNCZ7PSN1KOkYjao3QBoKDyn/jD57wbv
Uf3N245gf55BAj38LCe64O3gXbgT4TNiCWeRMMJ5gPuyF4QYPzXudDEQ8vyc9+c=
=4e0U
-----END PGP SIGNATURE-----

Re: proposal: add "packagers" field (list of strings (names)) to package definition

[Ricardo Wurmus]

Andy Wingo <wingo@igalia.com> writes:

>> Aww, too bad. I think it's a missed opportunity to be inclusive and
>> generous with attribution which might motivate people to contribute.
>> Guix is in a unique position with this opportunity.
>
> If there are other ways we can recognize people's work, we should
> consider them :)  Think of it this way though, a packager is a person
> who packages.  The value is in the verb, not the noun; packaging is an
> ongoing process.  We should reward the process, via NEWS, and perhaps in
> other ways.

With every release the release announcement also includes a list of
contributors since the last release next to a number of their
contributions.  I think that this is a very public statement of
appreciation and quite motivating (because of the numbers, so you can
even be competitive if you want to).

I don’t think package objects are good places to attribute contributors
(all of them?) to a package.  They are usually invisible to users

> Besides, no one wants to be the person that removes a name from a list
> of a package's packagers!

+1 to this.  It would be so awkward.  And not removing names from
packages may result in a really long list, eventually, and removing the
names then would be no less awkward.

~~ Ricardo

Re: proposal: add "packagers" field (list of strings (names)) to package definition

[Ludovic Courtès]

Ricardo Wurmus <ricardo.wurmus@mdc-berlin.de> skribis:

> With every release the release announcement also includes a list of
> contributors since the last release next to a number of their
> contributions.  I think that this is a very public statement of
> appreciation and quite motivating (because of the numbers, so you can
> even be competitive if you want to).

Right, and it’s intended this way.

I’m open to doing more in terms of public recognition, but I agree with
others that the ‘packagers’ or ‘maintainers’ fields aren’t the right
place.

Florian, do you have other ideas on how we could publicly show our
appreciation of contributions?

Ludo’.

Re: proposal: add "packagers" field (list of strings (names)) to package definition

[Florian Paul Schmidt]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

On 12/03/2015 11:36 AM, Ludovic Courtès wrote:

> Florian, do you have other ideas on how we could publicly show our 
> appreciation of contributions?

Not right now. If any ever occurs to me I'll let you guys know ;)

Flo


- -- 
https://fps.io
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2

iQEcBAEBCAAGBQJWYLGKAAoJEA5f4Coltk8ZIxsH/2QbTz9Q+WxrwyrYd4GuJmt9
9xU/wnMJQ3VXOaiYpLuEY9QSYlRPgHiTa3FdVuiyXTAC7lrNYR6v+Svg/Vh0dc/D
CG5JowIdrFH60PlFBtbe44WJn86FaSdAYVojjyzCn1SSjTQI/S6UWUHARyYrQGTA
SpFtgT57xXeUkPMQSG4Q85xDqA/P9xORZXRYHfGkYRAEItJegHIzje+Y/K5svEAW
dFhGX01NWMbjgqLeiruKBbX07FnNK+uN+hXkMDVkQJgZ7DOvTK8j2GZJvTu4v2CQ
1JT5Lu8uOyVPjMnhxg7xgrbvz1afuG0GAYguF0q0D/WJZFpQqQHgAocbputU0ws=
=Go/M
-----END PGP SIGNATURE-----

Re: proposal: add "packagers" field (list of strings (names)) to package definition

[Andreas Enge]

On Wed, Dec 02, 2015 at 03:59:52PM +0100, Florian Paul Schmidt wrote:
> Aww, too bad. I think it's a missed opportunity to be inclusive and
> generous with attribution which might motivate people to contribute.
> Guix is in a unique position with this opportunity.

The problem with attributions is that it is also exclusive of others who
might contribute to the package otherwise. It can create tendencies to
"own" packages, which creates a barrier for others to modify the package;
so inactive maintainers can become a big burden to the progress of the
project as a whole.

On the contrary to your statement, I think that Guix is extraordinarily
inclusive and welcoming for contributions; newcomers may immediately suggest
patches to all existing packages.

Andreas

Re: proposal: add "packagers" field (list of strings (names)) to package definition

[Florian Paul Schmidt]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

On 12/02/2015 05:50 PM, Andreas Enge wrote:

> The problem with attributions is that it is also exclusive of
> others who might contribute to the package otherwise. It can create
> tendencies to "own" packages, which creates a barrier for others to
> modify the package; so inactive maintainers can become a big burden
> to the progress of the project as a whole.

Um, it's a list that can have more than one entry :) If 300 people
touched a particular package definition, why not have 300 people
listed (if they felt inclined to add their name)? :) Also note that a
packager is very different from a maintainer.

> 
> On the contrary to your statement, I think that Guix is
> extraordinarily inclusive and welcoming for contributions;
> newcomers may immediately suggest patches to all existing
> packages.

I didn't say it wasn't (though I do have some qualms with the mailing
list workflow which might scare quite a few people off - especially
people that were born in this millennium, aside from email being a
fundamentally broken technology anyways - I'm an old fart that's used
to them and even I was weary to sign up to yet another ML). I was
saying there's a unique opportunity of it being even more so :)

But it's just a proposal to ponder a bit.

Flo

- -- 
https://fps.io
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2

iQEcBAEBCAAGBQJWXzvCAAoJEA5f4Coltk8ZTAQH+gJX0Rawc4wkItYwpQmPRKFh
zeupF9EXcRAHIGDNe0UYI0uJNQVmnIosqJYHowm/9tAXcueCsYQuemrF051XQdUI
Afby4FHIkfux6x8Exg4VygAxkOpYPfO5rxrdgtkJx8pMsGvkYGmpjOGn6Qjev+um
apL2QgnO6pw1AktL+ilxqepHV4j88J5rfPRfQqdh8RqftMvLHxtt9nXeJNbjRtXX
eWRU8TTppu+o/hxKKPy/1OZO3HT7FF93qj1hd/IzYqL8i8iiVy+Cgvcsgy1cW4Zo
cKQu1NlEYcvoOrLdN8Ni94j5u2wgGcKfeHZl2L1iIX9XECNbL3E/axEeCNjiOCA=
=IKVx
-----END PGP SIGNATURE-----

Re: proposal: add "packagers" field (list of strings (names)) to package definition

[Pjotr Prins]

On Wed, Dec 02, 2015 at 03:59:52PM +0100, Florian Paul Schmidt wrote:
> Aww, too bad. I think it's a missed opportunity to be inclusive and
> generous with attribution which might motivate people to contribute.
> Guix is in a unique position with this opportunity.

Git history is conclusive. There is no stopping anyone mining and
creating outside meta-data.

Pj.

Re: proposal: add "packagers" field (list of strings (names)) to package definition

[Florian Paul Schmidt]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

On 12/03/2015 05:46 PM, Pjotr Prins wrote:
> On Wed, Dec 02, 2015 at 03:59:52PM +0100, Florian Paul Schmidt
> wrote:
>> Aww, too bad. I think it's a missed opportunity to be inclusive
>> and generous with attribution which might motivate people to
>> contribute. Guix is in a unique position with this opportunity.
> 
> Git history is conclusive. There is no stopping anyone mining and 
> creating outside meta-data.

True, except for one scenario: If someone proposes a package not as a
diff with authorship information, but just as a text via some other
channel. Or maybe just a good suggestion which someone that provided a
diff might deem worthy of a mention :)

But many roads lead to Rome, every approach has pros and cons. I'm not
pushing this any further

Flo



- -- 
https://fps.io
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2

iQEcBAEBCAAGBQJWYLDVAAoJEA5f4Coltk8ZLm0H+wV3aNLTbEWU7d1qXSI7O1Is
9Ztc582iTQBanvTbzE3VqvCAGS/znF7gcjja1zM2CMZC5CGk5El5Pdtxno4n9EPs
Q7FDQ/NQiLqumC59EV6absyJ+hlCGQ0ySFeeCasdgDzoTMxKGXBJuZBRawsv1Gmc
TWkYf7VLNFA46JPtVWroqcPiGw0M/FjC80ztTk9hFsNf+lrFARqJSx47CbGxaTXD
PmhB8kwEqpILzREdD4s+JO+Hw5RCK3kCGVyqzUaomlwgqaZJHZEkIc21vtkPbz26
L8TB4idnEQFgr7rGtqfErYOSH6cNxYrlxVJjYboMIO/BlzynfwWvCTh3gCFRPs0=
=Agdj
-----END PGP SIGNATURE-----

Re: proposal: add "packagers" field (list of strings (names)) to package definition

[Ludovic Courtès]

Andy Wingo <wingo@igalia.com> skribis:

> On Tue 01 Dec 2015 08:12, Florian Paul Schmidt <mista.tapas@gmx.net> writes:
>
>> ...and encourage its use. The intended semantics is to list people
>> that have contributed to the packaging effort. The motivation behind
>> this proposal is that in many free software projects attribution can
>> be a major source of motivation to get people involved. Having the
>> packagers be first class citizens in the package definitions (as
>> opposed to the information being only implicitly available through
>> e.g. "git blame") would allow things like "guix package" or the
>> package list on the website to display the contributor's names.
>
> Since you ask for opinions, -1 from me :)
>
> Currently Guix packages are more-or-less collectively owned.
> Introducing this field implies to me an introduction of ownership of
> packages.  Ownership has a number of negative effects: it can inhibit
> casual fixes and it can introduce unnecessary conflicts.

I agree.

For a while I wondered whether attaching names to packages via the
‘maintainers’ field would help people feel responsible, and thus help
keep packages maintained (but obviously this could only be used once the
ratio of the number of contributors to the number of packages has become
reasonable.)

I’ve come to the same conclusion as you: that collective maintenance is
even better, and actually works reasonably well.  The whole distro is
our commons.

Cheers,
Ludo’.

Re: proposal: add "packagers" field (list of strings (names)) to package definition

[Taylan Ulrich Bayırlı/Kammer]

ludo@gnu.org (Ludovic Courtès) writes:

> For a while I wondered whether attaching names to packages via the
> ‘maintainers’ field would help people feel responsible, and thus help
> keep packages maintained (but obviously this could only be used once the
> ratio of the number of contributors to the number of packages has become
> reasonable.)

I actually found myself wishing for a way to list all packages I
packaged, so I can update them (say, somewhat obscure and possibly
indirect dependencies of packages like mpv, which might end up very
out-of-date since nobody directly cares for them) but then I remembered
this is easy to do via git log grepping, especially thanks to our strict
commit format.

    git log --author=Taylan --grep='^gnu: Add ' --format=oneline |
      sed -rn 's/.*Add (.*)\.$/\1/p'

Taylan