[Mageia-dev] RPM5 AND MAGEIA

[Per Øyvind Karlsen]

2011/3/2 Dexter Morgan <dmorganec at gmail.com>:
> On Wed, Mar 2, 2011 at 4:21 PM, devzero2000 <devzero2000 at rpm5.org> wrote:
>> Hi to Mageia
>>
>> I am part of rpm5 (but I follow rpm from more than 10 years) and
>> recently I did join Mageia, although I almost always used for work
>> other distros and OS (AIX, SVR4) that I also deal with Mandriva and
>> security (CEH) https: / /
>> www.redhat.com/wapps/training/certification/verify.html?certNumber=805007215126804&verify=Verify
>>
>>
>> Apart from the rest - of which i will ask for sponsorship when it will
>> be - I wanted to know if there are plans to move to rpm5 by Mageia,
>> such as Mandriva has been doing lately. Rpm5 already has a builtbot
>> with Magela and rpm5. I can, if you can think useful or have plan for
>> this, lay the necessary modification to enter into rpm5 Mageia, with
>> the features of Mandriva cooker - fingerprint, syslog, etc. without
>> trademark ecc- and produce a first rpm rpm5 for mageia , which also
>> contains the functionality required by the passage to the "RPM
>> ACID " feauture (berkeley db conversion)
>>
>>
>> is there interest? Then if someone wants to directly participate in
>> rpm5 I think that there should also be no problem.
Exactly, opposed to rpm.org upstream which only has a few redhat employees
working on it with no public communication, coordination and discussions between
distributions, rpm5 is community driven and has the active involvement
and more open participation from several vendors/distributions on different
platforms..

Your (Mageia;) participation is more than welcome and active direct involvement
in upstream development is generally no problem at all in arranging. :)
>>
>> Opinions ?
>>
>> TIA
>>
>> Regards
>>
> Hi,
>
> Please read the archives we already talked about this several times on the ML.
I saw several people talking and asking about rpm5 on both mageia-discuss@
& mageia-dev@ list, I even myself (and also relayed Jeff's) offered to maintain
and provide any help you'd accepted on it for Mageia a couple of ties without
any response. I saw several people asking semi-frequently asking about it on
different occations since, but without any actual discussions considering the
topic taking place.
I can only recall seeing one post from Thierry to mageia-devel@ where he
made a suggestion for rpm.org, but never saw anything about it since before
the upgrade to newer rpm.org version was done by someone I'd never seen
the name of before except for a fictional character on a TV series probably
totally unrelated(?;) and even much less so ever saw participating in any
of the discussions concerning it previously.

But I only follow these two lists, there might be other lists I don't know of
(ie. communication around cooker tends to go across and between
maintainers at mandriva.com (that a lot of people don't even know of) and
cooker@ lists, and then also at some various other internal lists at Mandriva
that's restricted, so for all that I know the discussion could've taken place
somewhere that I might not even expect to be open for me;) and where the
actual discussion took place. I dunno, do you have any references to the
discussion? I'm usually quite interested and curious about such discussions
in general. :)
>
> We are using rpm.org for now, and we are happy with this, we are
> already working with rpm.org teams and this works pretty nice.
Telling that you're "already working with rpm.org" and being able to comment
about the experience less than a day after making first post to rpm-maint list
introducing yourself and Mageia seems a bit premature to me, but it might
be a cultural thing.. ;p </troll>
>
> But you can go and help mandriva as they have several problems with rpm5.
Yes, and we're even actually aware of these ourself (*coughcough*;), currently
none of those (reported) is of particularly significant size or causing major
headaches to people in general for a little while now though. There's still
some few issues left that *must* be worked out in time for 2011.0 none the
less though. The number of other issues, room for improvements, planned
features, cleanups, ideas and what not otherwise is rather long and rather
non-finite and not likely to ever be finished and really "done" with,
but at least
I can say that it's significantly shorter now after changing to new rpm version
with a lot of the more long-time work already done in it before switching. :)

I'm more than willing to accept (and provide) any help, interest and discuss
any details concerning rpm, regardless of distro and rpm version/fork, be
it related to any of current issues affecting distro, other things I have on my
TODO list, or other aspects and focus independent of these. I'm very open to
try my best to scale and allow for others that have the interest and skills to
discuss and contribute, either directly or indirectly!
>
> In addition, rpm5 ( i hate this name as this makes think rpm5 is a new
> version of rpm which isn't true ) is just another rpm and i think that
> this is better to stay compatible with the most distributions (
> fedora, opensuse, ... ) instead of using rpm5.
You claim for it to be old?
And by what is your definition to "stay compatible" with most distributions?
Are you referring to different distributions doing releases,
or compatibility between individual releases of the specific distributions?

For compatibility between different distributions, the only
distribution actively
involved with actual upstream development of rpm.org is Fedora, with
complaints heard from others on despite it being over a year since Fedora
& OpenSuSE had their RPM Summit (which "only" covered rather "smaller"
issues concerning packaging policies than involvement and active
participation in
upstream development), nothing has been heard between the two since..

The amount of changes from upstream carried in OpenSuSE's rpm was
quite huge and making incompatibilities between the two bigger than between
ie. rpm.org & rpm5.org, with the situation not changing much since from last
I heard.. And considering that I've fully reviewed and merged all patches in
Mandriva dating all the way back to 1999 in rpm5 upstream along with a lot
of practices, macros, scripts etc. as well in both directions, level of
"compatibility" between Mageia & rpm5 is far greater than any other (with
other rpm5 based distros being more compatible with Mageia than Mageia
is with non-rpm5 based distros as well;)..

Even between individual releases from the individual rpm forks there are
always similarly incompatibilities often happening and impossible to get
around while keeping full "compatibility" intact.
For the upgrade in cooker to rpm 5.3 there was two bigger changes happening,
the first one being the rpmdb changes requiring a conversion to take place,
this was inconvenient but a proper and safe procedure to achieve it
was mostly worked out before hitting main/release, with the rest of it
pretty much fully perfected since (with support for conversion back and
forth added as well to achieve "compatibility"), making this issue only
temporary and not really of much concern outside of that limited period
(except for when upgrading from cooker to Mageia, I've again given
pointers, suggestions, specific directions and even offered to provide
help on doing it, but as usual and to little surprise, interest was close
to zero..).

The second issue causing bigger incompatibility issues and pains is the
new distepoch tag I've introduced in Mandriva and that we're the first
real adopters of (it's not even yet officially supported upstream).
This is more of a particular feature implemented by myself rather than
so much related to the specific rpm version beyond the rpm version
I used to implement it in first (I have it on my TODO to port to rpm.org
later and has discussed it with positive interest from upstream of in the
past as well btw. ;), and also causing greater incompatibilities and
headaches between tools and the packages built for the specific
distribution. Worst issues has been worked out for this one as well with
various related fixes to deal with incompatibility being ported back to
stable releases along with some some other unrelated fixes as well
(if interested, you'll find the 3.37.2 branch of perl-URPM in mandriva svn,
which except for the fixes mentioned is basically identical to your 3.38
inn Mageia).

These are anyways not issues impossible to manage, and much of
the intent behind distepoch, ironically enough, is rather to improve
future compatibility with other distributions by moving distro specific
stuff out of the src.rpm. ;)
You'll always keep run into these issues, neither direction attempted will
free you from ever running into them now or in the future.
If you really wanna discuss "compatibility", you *really* need to define
what is even somewhat implied by this as *compatibility* isn't something
you'll be able to achieve nor anyone else really has that much interest
in chasing..

I honestly don't even expect you to read on and all the way down to even
here to begin with. If you do, good, it's one step in the right direction, if
not and not really grasping these issues, you're doomed at even
attempt starting to begin with. :p

So first of all and most important to begin with is how you yourself wants
to deal with the particular issues for your distribution, how to efficiently
solve your own individual problems to  begin with, how to improve things
to make life for anyone involved in maintaining the distribution currently
easier.
Then you can start think about "compatibility", first of all between your own
individual releases and tools (independent of upstream) to begin with and
concern yourself on really solving things the "right way".
Then on you can finally start putting thought into "compatibility" beyond
own distribution, how to best achieve it, both on your and others' end.
With different tools in use and development, one true answer and solution
cannot easily or even always found at all, and the debate about what being
upstream or not, what's limited to your own distro only or not and what not
gets murkey and the choice needs to be done from not so much based on
which fork (or even tools to use) comes from who or what or anything, but
rather based on what's most feasible to you & maintenance, to what extent
gains in return you expect, existing involvement, knowledge, insight and
actual skills on achieving the goals set.
And uhm, yeah.. In the end it gets complex and not strictly just about
making a simple choice expecting great results in return from that alone
and sticking to it..

Tsktsk, my post probably ends up being labelled as flamebait, troll, too long
and boring, biased and what not.. I just felt like giving what was intended
as a somewhat brief answer in response to a lot of nonsense and FUD
(mainly built on ignorance and power of suggesion), and also from greater
interest in the particular topic you bring up, but turned into a longer
semi-rant(without even straightening out half of what I intended;) though.
Appologies given upfront!

So I'll try avoid getting deeper into further details. :p

Again, main point is that there's more to it than anything I've seen trace
of being discussed or anywhere related to what you say you'd like to
achieve. As "not following mandriva, maintain everything alone, follow rpm.org"
doesn't really answer it much, but rather raises more questions and concerns
over how you intend to succeed (but I honestly don't know you at all,
so you might be full of nice surprises:) and with whom.
I'd be happy to try help the best and be able to make life easier for everyone,
and I'm more interested in this than on what specific distro nor rpm fork
you're using, just as I'm convinced and sure the importance lies much
more in "how to" rather than in "how you're not".


"Compatibility" between the two different rpm forks solely alone on their
own independent of distributions is rather trivial to achieve (and usually
already done by both upstreams) and honestly not that big of an issue,
as for how to deal with differences between the individual distributions
gets far more complex again..

It's a topic I'm actually doing parts of my thesis around, so if you have
any specific thoughts, ideas, meanings, questions etc. I'd be very
interested, and very happy if finding interest in discussing and working
together with others on. :)
>
> This is just my opinion but i would like to not see rpm5 thread again
> and again, that start for my part to become boring.
If you keep seeing it again and again, maybe it suggests that the
discussion never (sufficiently) took place to begin with and an actual
consensus known to everyone and what based on might be missing
in order to reach any finality on the topic? ;)

I've only seen threads with questions and quite superficial discussions
not really answering many of them, but it might be my personal
bias affecting critical sense and standards too much.. :p

Oh well, tried my best being constructive and genuinly helpful,
expecting nothing, but would be positively surprised and happy for
any serious discussion and interest, if discovered!

Need some sleep to achieve greater consistency, focus and coherency..
--
Regards,
Per Øyvind