Lars Müller wrote:

On Wed, Aug 10, 2011 at 02:10:26AM +0200, Joachim Schrod wrote: [ 8< ]

...

(I need to patch 4-5 init.d scripts in each openSUSE release; let's hope systemd improves that.)

Are these generic changes or are these jschrod only changes?

Lars, that's hard to tell. If I would be sure that they are generic, I would have filed bugs already. Almost all of them revolve around our usage of autofs. Besides specific usages like "apache document root on automount NFS", perhaps the most generic usage is home dirs via automount. There, some systems, e.g. postfix, need to wait for autofs to be able to process user files like .procmailrc or .forward properly. If the soft dependency to autofs is not there, they will reject emails to those users during startup. Another generic issue might be that ypbind doesn't wait for ypserv. But then, I know that usage of NFS home dirs and NIS (on some legacy sub-networks of big clients, FWIW) is not exactly the target group of openSUSE. Thus, we don't bother with bug messages and establich patching the issues as part of the deployment process. (Frankly: I'm the CEO/CTO of a small technical company. I only enter bug reports if I think there's a chance that they will be handled. Otherwise my time is better spent on my own Open Source activities. (E.g., for the last two weeks, I have been the sole CTAN admin not on holidays. For me, TeX Live et.al. are more important than openSUSE bug reports. ;-)))

If they're generic please file bug reports and attache your fixes.

As I wrote, I don't know if they're sufficiently generic to get enought developer attention to get fixed.

If they're are special purpose consider to suggest a way to make you feel more happy by adding a hooking facilitry which allows you to add your special needs.

Actually, I hope that systemd will improve the situation for us, and -- more important to me -- for several of our clients. Dependency specification is more sane there, in theory. Still, there remains the issue how one can specify local dependency changes in a supported process. Lennard, Kay, et.al., are supportive, but are also very harsh to brush away non-standard environments as not relevant. Let's see what direction that will take. Since I don't want to catch flames for this comment, let me emphasize that I don't have problems with the systemd development process. If my customers don't like the direction that's going, we can fix that ourselves, just as we do since decades. In my company, we do data center planning & support as part of our business, and I see *much* more non-standard environments as I would like -- and as is sensible, to be frank. Well, it will be *very* interesting to see how this will pin out; if the my-laptop-is-all-the-world-centric developer crowd will have to face a backslash from the servers-are-all-the-world admin crowd or if they will push their changes through. (Personally, I think that both are wrong.) Currently, it seems to be constructive, witness Frederic and Wolfgang on the sendmail startup definition. I didn't even know that there are systemd generators. ;-) So, I have great hopes for that direction. Joachim -- =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Joachim Schrod Email: jschrod@acm.org Roedermark, Germany -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org