Re: Bug#797473: frama-c: New upstream version
Hello Mehdi,
Le 2015-08-31 09:09, Mehdi Dogguy a écrit :
Debian Jessie has been released during end of April 2015. Updating Frama-c
before Jessie's release was not an option. We could have uploaded a version
in experimental but we preferred focus our energy on getting the Jessie
release out.
The current release-every-packages-at-the-same-time approach is, in my
humble opinion, part of this issue.
Frama-C uploads were done rather quickly in the past after a new upstream
release.[...] I originally
planned on updating Frama-C during this summer, but various factors led
me to focus on other areas (more interesting for me).
I have never and will never criticize a volunteered based work, I know
well enough its consequences and limits.
I understand this remark was meant as a constructive criticism, but I
really wonder if it serves any positive goal.
Certainly not for this Kurt's issue and probably not enough for the
overall issue. At least I wanted to give a feedback (which in my humble
opinion is always wanted) which of course you are free to ignore.
Debian provides a highly
customizable high quality platform.
A (really!) naive question: how Debian can be customized? I always saw
it as a monolithic platform, tens of thousands of packages rather well
assembled. Of course, one customization dimension is the ability to
install a subset of those packages, but all Linux distributions provide
that, no?
Users needing a stable platform
with very latest versions of some software should invest some time to
make that happen. This contributed to the fact that Debian has the highest
number of derivatives. Sometimes, a new distribution with a more up-to-date
software stack pops up. But this doesn't come without making some
compromises.
You're of course right. And I am well aware of my contradictory
requirements: a very stable platform on the one side and latest version
of *some* software on the other side. The "some software" being
different for each user.
But in my opinion, this is part of the issue: with so many packages, it
becomes more and more difficult to release all packages in acceptable
state (from freshness, stability and security points of view) at the
same time. As a user, for a long time now, I had the feeling that the
criteria were not reached: outdated software and not so good quality
packaging (e.g. some functional bugs are not fixed in the package while
fixed upstream because "only security or critical issues should be
fixed(tm)").
I now think you need a way to accommodate several versions on the same
piece of software at the same time, and combine them in any way the user
wants. I obviously understand the difficulties of such a target, but at
least this target should be a long term objective (and it might be
reachable with adequate tools, rules, new development and research).
Also, as you pointed out, Debian has some means to get some packages
updated for a stable release through the Backports mechanism. But this
also requires some efforts. Note that anyone can make backports and publish
them for the common good. It doesn't have to be made by the package
maintainer.
I didn't know that.
As long as users (with specific needs) keep waiting, you can be sure
that nothing will happen. At least, that's my understanding of Debian's
development model and that's what pushed me to get started and
contributing. And this is not even specific to Debian, but also true
for the wide majority of FLOSS projects out there. The fact that Fedora
has an up-to-date Frama-C package is only explained by the fact that there
more motivated people (or person?) to keep packaging Frama-C.
You are right of course!
If you are concerned about the updateness of the OCaml packages in Debian,
maybe you could join the team and help? I am pretty sure this will be more
effective than pointing users to Opam (which is not made for people looking
for a stable platform, but for developers) or Fedora (which is a completely
different system).
I still consider that my answer was practical and correct. As I said, at
work we are using opam to make Debian based virtual machines with latest
OCaml tools with no major issue so far. And some users might change
Linux distribution according to their needs.
Now, to answer your invitation to participate:
* My first and easy answer is: "I unfortunately don't have time". :-(
With two young children, I lost all my ability to contribute to Free
Software. That will hopefully change with time;
* I'm not very interested in Debian packaging, despite having a friend
pushing me to do it for at least ten years :-) and having looked at some
documentation and doing some attempts in the past. And having read this
mailing list for years, the infrastructure and roles seem still *very*
obscure, the use of acronyms being only rivaled in network standards; ;-)
* More fundamentally, I am no longer convinced that the technical
solution of current Debian packages is the good one (with the disclaimer
that I probably don't know enough all the details). It is only possible
to install one version on one package at a time and the packaging task
seems (to me) very manual and fragile (e.g. the ability of a package to
put its files anywhere on the file-system, no well defined interfaces
between packages or sets of packages, no systematic test or formal
verification of assumptions between packages, ...). As far as I know, it
would be very difficult to change that in Debian. :-)
I don't want to appear to pessimistic. I am very fond of Debian
independence of any company, its focus on Free Software, its democratic
internal working and several recent changes (e.g. Zack et al. work on
package dependencies and systematic checking, openness to non developer).
Hopefully, I'll provide in the future more concrete contributions to
Free Software that discussions and ideas on a mailing list. ;-)
Thank you for having taken the time to read and answer me.
Best regards,
david
Reply to: