On Fri, Feb 28, 2003 at 07:13:35PM +0100, Stefan van der Eijk wrote:I'm looking mainly at i586 and alpha. I'm peeking at ppc and ia64 at the moment. Trying to get them in sync.
I rsync (over ssh) directly from kenobi. The mirrors are too slow and too f*cked up for the things I'm doing. I'm happy that I'm allowed to use rsync on kenobi, if that is taken away that will mean the end of me uploading the alpha port.
Well my personal mirror which I get off carroll is fine for i586. It's just ppc that's messed up and well it seems to always be that way. I don't mess with the other archs so I'm not sure about them.
uh.Yup... a lot of frustration in the air. Many people have the same feeling. I'm just wondering WHY?
What's wrong with mdk?
* Is it a controll / trust issue (scared that contributors are going
to f*ck up their product)?
* Is it a resources problem?
Neither. It's a communication issue.
Been there. In the end I'll force them through myself. But not everybody has those priviledges.a) On several occasions I have felt as though I have been responded to in a rude or disrespectful way after contributing something. This does not instill a desire to contribute, especially in a situation where feeling good about what you did is the primary benefit of contributing.
b) Having my contributions flat out ignored.
Well, if it ain't documented, then I guess you can as much as the system will allow you to do...On many occasions for packages in main (and when I didn't have write access to contrib) my submissions of fixes to packages would go unanswered for long periods of time. Stew finally put through wireless-tools because he got tired (I guess) of hearing me complain about it.
c) As you point out below the documentation on what the policies are is severely lacking. I don't really know what I'm allowed to do with my write access to contribs.
Been there with Laurent a few times. At the moment I'm not touching kde packages because of this.Several people (e.g. Oden) have run into situations where people feel like their toes are getting stepped on.
and exceptions, which aren't documented either.Not to mention some policies aren't clear at all so there are a dozen different implementations.
And some of the documentation that exists is flat out *WRONG*.And / or outdated.
I'd merge everything into one repository. At the moment keeping the boundry between the repositories in check isn't being done properly. Quite often (Build-) dependancies cross the border, which is bad.d) There isn't a single way to contribute. There really should be one concise way of contributing packages. I.E. club and contrib probably ought to be merged.
IMHO quite some packages in cooker have the same importance as some contrib packages. Just let the mdk people maintain the core packages, and let the contributors do the rest. The number of core packages should be about half of number of packages in cooker at the moment.
Yes.e) Many decisions are made without consulting or even explaining the rationale behind them to the community. When asked about them many times you get silence or a flat out nasty response. Perfect example is the licensing policies on MNF.
Yes.f) Many contributions are made overly onerous to contribute. If you aren't a packager it is difficult to get anything in. Mandrake makes no effort, IMHO, to foster contributions to manuals or other non-technical things. Someone did labels for CDs... and has seen Mandrake simply ignore these labels. (Though that discussion is active again on the club-volunteers list).
Well, at the moment their pretty much ignoring it... at least, what I'm seeing. In my daytime job (consulting, mainly for multinationals) Mandrake doesn't even come close to being on the list of products to use. Prefered Linux distro = RedHat. Next up would be SuSE.I totally agree. At the moment I'm willing to work on (the process of?) improving the process, not at improving the product. Once the process is right, then they'll have me onboard again to work on the product.
I'd also like to see more things documented:
* Like what contributors are allowed to do and what not. * How things are to be done. * update the rpm howto document
And have the management make decisions and PUBLISH them and stick to them. If the alpha port isn't going to be supported --> kill it.
*nod*
Get a real business model. More focus on the enterprise, less on end-(l)users. In the end, the enterpise market is where the real $$$ are.
I'm not sure that you have to do enterprise stuff to make money.
Distributed security stuff should get more attention at MDK. The market is asking for it.
Just wondering if desktop users are going to pay the bills...But I do think Mandrake needs to target its products to its userbase better. Right now its userbase is primarily desktop users.
For enterprises it's less of an issue. They are paying gold & diamonds to keep their prop. unices running. Basically the Linux enterpise products are mostly support offeringsThere is a lot of confusion about what Mandrake should be used for. That could be solved by making a "server" Mandrake branch like they have for firewalls. I guess Corporate Server fulfills this to some degree. But it's kinda pricey for a lot of people.
You're not. There are quite some people with the same feelings... At the FOSDEM in Brussels I had some discussions with other mdk people (some ex mdk, some plf crew), these people were saying the same things...I've suggested it before, maybe we (the mdk community) should write a document which reflects how we want things to work and make demands. Otherwise boycott contributing to mdk.
I've worked on something to this effect. But I'm so frustrated that I feel I'm alone in trying to get any changes effected.
OK. No problem. I'd love to review it.If you want I'll show you what I have, though I'll have to ask that it stay confidential for the moment as I want to finish it before releasing it to the world.
StefanIf you want to discuss catch me on the freenode IRC network. My nick is ResDev.
smime.p7s
Description: S/MIME Cryptographic Signature