On 2011-04-20 13:13:29 (-0600), Stephen Shaw
On Wed, Apr 20, 2011 at 12:02, Pascal Bleser
wrote: On 2011-04-18 11:02:49 (+0200), Andreas Jaeger
wrote: [...] The build service team has been developing a completely new proxy server that uses ldap directly to get rid of the broken ichain proxy. The systems is ready but we would have loved to give it some more testing and had hoped that ichain would be with us a few more days... [...] But it is yet another example of something being developed as an in-breed solution in a couple of offices at Maxtorgraben, 5 in Nürnberg. [...] I don't have a problem with how this was handled. I don't think it came as a surprise to anyone that they were switching. Where this was
It does come as a (very good) surprise, because we've been plagued by iChain since quite some time now. Now suddenly, we hear that there's a replacement. Don't get me wrong: it's great that there will be a replacement, I don't think anyone argues about that :))
something that was impacting a mission critical service opening it up to the community would have only pushed the fix off. I mean seriously. Look at any of the "way-too-many mailing-lists" and the discussions on them. It takes us weeks if not months to argue over the topic to only have everyone get upset and walk away from it. And heck, even on the occasion that we actually arrive at a decision, close the topic, effectively implement it we still have people rehash it over and over again.
I think that there is definitely a difference between technical topics and soft topics, for one. And then, there is rarely any discussion about such topics in the first place. It's "we've done this", at best, and that's it. Discussions getting too long is actually an entirely different problem, and a big one, which certainly requires attention and taking action, but that's not an excuse for not at least blogging or posting about something to give people a chance of chiming in with ideas and advice, or even with code. Heck, at least blog about it. "Michael and I have started to think about a replacement for iChain, because yada yada. There are a few options out there, and we need yada yada. mod_auth_memcookie looks neat, Michael is currently investigating it. If you have some experience or ideas to share, poke us on yada yada" How hard is it ? (I know, I don't do it all that much myself :))
So, the long and short of my rant. If getting OBS back to a stable state in a matter of days/couple weeks meant cutting out the weeks of arguing over what implementation would in theory be better, then I'll take the executed decision any day (when talking about getting a mission critical service back up and running correctly).
It's not about this particular case, but in general, those discussions never happen that way. And that way of working doesn't happen with non-mission-critical topics either.
Just as a side note... How many community people currently contribute to the development/design of OBS? I'd say with exception they are the only ones in any real position to give advise here.
Hey, *precisely* my point :) "Here is what we've done. No we won't discuss it nor change it. Now contribute." -- you really think that has any chance of working? I don't think so, personally :) cheers -- -o) Pascal Bleser /\\ http://opensuse.org -- we haz green _\_v http://fosdem.org -- we haz conf