Tilman Sauerbeck <tilman@crux.nu> wrote:
Matthias-Christian Ott [2007-10-05 17:51]:
Hi,
My first contribution to CRUX was the libarchive support for pkgutils (FS#121). I took a month to get this feature into the git repository. There were some trivial problems with libarchive, but it became apparent to me that CRUX had maintenance problems. I was not really happy about this fact, but I pretended myself that this was not a real problem.
I don't really see the issue with a non-critical patch taking a month to get merged.
I think patches should be merged as soon as possible.
But all these incidents never discouraged me from using and liking CRUX. I was not happy about the situation, but I neither wanted to fork yet another distribution nor I wanted to switch to another.
In May 2007 tried to integrate libtre in pkgutils without any success. This was not a matter of great importance to me, but I had real doubts about the organisation of CRUX from that on.
Did you see my comment on the bug tracker? I don't see the regex handling as performance critical, that's why I didn't apply your patch back then. Why didn't you provide profiles to convince me how much faster pkgutils/prt-get can go in typical use cases with libtre?
I did not provide any profiles, because I did not have any.
I tried to get a developer account and proper e-mail address on the CRUX server in order to be able to participate in development, maintenance and organisation, but nobody replied, my e-mails were rejected. I contacted Per and told him about my plans, but he could not really help me, because he was not a CRUX developer anymore.
Right. Who did you write to? I've never heard about those mails.
I wrote to crux@ and crux-devel@.
The next logical step would have been to write to the crux-devel mailing list, btw.
I did so, but both messages were bounced, because I was not subscribed to the list. The e-mail to crux@ was reject by the administrator.
it is just a matter of time. CRUX lost its importance, its fans and community. It is hierarchically structured and slowly developing.
It always was, I don't see the regression here. In fact it's much easier to contribute these days than it was "back then".
But this does not justify it. Just because it was worse in past it is not a good thing nowadays.
When I recognised that the C++ version of pkgutils was going to be replaced by a C version, I knew that this would have far reaching consequences for the design and philosophy of CRUX. I think it is not longer the great design philosophy of Per Lidén; it is not longer the thing I liked about CRUX.
Huh? Sorry, I don't understand why the pkgutils move is such a big deal for you. IMO it doesn't affect the philosophy of CRUX at all.
Yes, but it affects the way the philosophy is implemented and expressed.
I thought about the future of CRUX and conceived some ideas and demands:
discussions, development teams, ...) * open ports system (easier updating, submission and maintenance, open to anyone) * entirely public wiki * (perhaps) merge with CRUX-PPC
* automatic version, standards and integration tests for Pkgfiles * more precise and stricter packaging standards * no more footprint mismatches (isolated builds, USE flags like system) * complete localisation (no --disable-nls, ...) * (simple) Pkgfile standard library (common mirrors, functions) * multiple architectures
* improvement of pkgutils (attribute, hooks/events, pkgutils library)
I thought you hated the pkgutils work I'm doing?
"Hate" is something different, but I did not like it very much. You changed something that was already good in favour of something that is not better.
Some points on your list are totally NOT compatible with Per's idea of CRUX, I think (like NLS, USE flags, ...).
Localisation is compatible if it is optional or removable via hooks/events. USE flags (better: feature flags) are a consequence of the software that already violates the philosophy of CRUX and UNIX. CRUX has to find a way of dealing with them and feature flags are a quite simple way if they are not abused.
And I'm sure there's lots of users who _like_ not having NLS and no USE flags. Really :)
If it is optional, they will be fine with this.
Seems you want Gentoo with ports/pkgutils :)
No, Gentoo is bloated. It has several design mistakes.
Let us (the community) take over the development of CRUX! We are the people, we can change our distribution! The future of CRUX is up to you! You can decide on your own!
That's kind of rude considering we're still active. Feel free to fork if you hate CRUX though.
I prefer co-operation to competition. But there is no alternative, I will perhaps fork. -- Matthias-Christian Ott