Re: Debian Math Team
Hi Andreas,
Thanks for finding time to reply to my worries.
On 2021-11-01 16:19, Andreas Tille wrote:
> 1. Another "one-man" team
> I think Ole gave a good answer here: Its not about creating a
> one-man team but rather attract more people to the team - from
> inside and outside Debian. For those who have any doubt that this
> can work: There are 23 people who confirmed, that they became
> DDs *because* Debian Med exists[4]. Debian Med is now nearly
> 20 years old. I would love if Debian Math would beat Debian Med
> in attracting new developers. (Andrius, you even belong to this
> set. ;-) )
It is true that I am among these 23 :) My reason for being in the list
is mostly pragmatical: it was Debian Med people who signed my key and
advocated me to become a DD, and for this I am very grateful. I believe
being very open and very friendly to newcomers is one of the strongest
sides of Debian Med.
> Those 20 years when the Debian Med project have teached me that
> it is very important to advertise the fact to the world that Debian
> is targeting specific fields of science and IMHO mathematics is
> a) well worth to be advertised
> b) has lots of technically competent people beeing potential DDs
>
> 2. Further fragmentation of debian-science
> I do not think that another Blend will lead to fragmentation of
> Debian Science. Sure, some mathematics related packages will be
> moved sooner or later but I personally can not see in how far
> this might weaken Debian Science. I personally see also additional
> contributors for Debian Science once more mathematicans might
> join Debian (as I'm very positive about).
>
> Am Sat, Oct 30, 2021 at 06:18:28PM +0530 schrieb Nilesh Patra:
>> Hi Andrius,
>>
>> Thanks for replying. See below :-
>>
>> On Sat, Oct 30, 2021 at 10:33:02AM +0300, Andrius Merkys wrote:
I am skipping most of your replies to Nilesh's response to my mail. I
ACK the skipped parts.
>>> Furthermore, from my experience one does not need domain
>>> knowledge to successfully package and maintain packages in Debian.
>>> What makes more sense to me, is organizing packages into teams based on
>>> programming languages and build/test systems, as such teams indeed
>>> possess specific knowledge. I think most of the mails asking for help in
>>> debian-med concern language and build system problems, not
>>> domain-specific issues.
>>
>> I'm sorry, but I have to admit this argument of yours is sloppy, Andrius.
>> By this logic, we could push entire debian-med python packages into python-team,
>> java packages into java-team and so on...
>
> While this would work in principle the point of creating a Blend is to
> attract experts who know the algorithm and internals of a certain
> software to craft sensible packages and enable thorough testing. This
> is usually not simply a programming language issue. In Debian Med we
> have several upstreams maintaining their software as Debian packages
> (after sufficient teaching of the packaging process and for sure
> sponsored by a DD). IMHO exactly this is the strength of the Blends
> concept to pair experts of the software with packaging experts. Even
> better if this can be completed to involve users into that effort which
> does not (yet) work as good as expected (but this has IMHO more external
> reasons which we can hardly solve).
I see your point. However I am not entirely convinced that teams are
essential here. Maybe separate mailing lists could be enough? In the end
upstreams mostly work on one-two source packages, and even if they
become DMs they do not get push/upload permissions for all source
packages of a team, do they? Domain-specific mailing lists could indeed
be a go-to place for newcomers, those in need of help with packaging and
sponsoring and so on.
>>> I am worried reading about R packages being moved from debian-r to new
>>> debian-math. I am afraid doing so might negatively impact their quality.
>>
>> You are right in your worries, but I have some statistics to present here.
>> See here[1] or more specifically, look here[2,3]
>
> I'm not worried about moving R packages to some other Blend. It has
> turned out to be a good idea to move all R packages from Debian Science,
> Debian Med Debichem and possibly others into one language specific team.
> However, this had happened since R packages are extremely uniform and
> usually come with test suites that can be re-used which to some extend
> is taking over the role of an expert knowing the software. There are
> also not really any specific decisions to make about the packaging since
> everything is really straightforward.
>
> This is absolutely different to software written in Python, Java or
> anything else.
I disagree. I find at least JavaScript and Perl packages quite uniform,
and I have an impression that at least Perl packages outside the Debian
Perl Team are generally in poorer shape than those inside the team.
>> The number of pure math software in R package team is in no way overflowing, so I really think this should
>> be manageable. The probability of it having a bit-rot will be less -- atleast not high with me, Andreas, Doug et. al.
>> being around.
>>
>> However if you very strongly feel about it, we could leave the R packages where they are and continue maintaining
>> them under R package umbrella.
>
> I'd say any Debian Math member who is interested in R packages should
> simply join pkg-r team and be done.
Agree.
>> [1]: http://blends.debian.net/liststats/
>> [2]: http://blends.debian.net/liststats/uploaders_r-pkg.png
>> [3]: http://blends.debian.net/liststats/commitstat_pkg-r.png
>
> [4] https://wiki.debian.org/DebianMed/Developers
Thanks again for discussing this with me!
Cheers,
Andrius
Reply to: