My personal pkgsrc FAQ

[Hasso Tepper]

Because I happen to be in the position where my person can be associated 
with pkgsrc, there are lot of questions asked from me whether and how 
situation with packages can and/or should be improved etc. I'll try to 
answer most of questions in this mail (just my personal FAQ) and to start 
some discussion.

Q: Pkgsrc sucks, why we don't use <some random> package management system?
A: There is no real alternatives. Unless someone finds 10+ people working 
   on our own packaging system (packages), there is no point to have "our 
   own" system. And unless someone finds other packaging system with focus 
   on portability and with reasonably big number of packages, there is no
   point to use something else either.

Q: But I'm willing to help to port <some random> package management system 
   to DragonFly!
A: You can do whatever you want with your free time, but unless you have 
   something done which can be really used, don't expect anyone in the 
   project (and especially me ;) help you.

Q: Can you do this and/or that (pkgsrc related task)?
A: In general no. I just happen to be a pkgsrc commiter. I do regular bulk 
   builds and try to take care that pkgsrc doesn't suck _too_ much in 
   DragonFly, but that is pretty much all. It is already more than I'd 
   like to do in this area.

Q: Can you fix <some random> package?
A: You can try to poke me, but don't expect it to be too successful. If I 
   think that it can be useful for me, then yes, but anyway try to fix it 
   yourself and submit the patches to NetBSD GNATS. 

Q: You do bulk builds, can you make packages from these available?
A: No. These builds are actually irregular, are often incompatible with 
   each other (I play with defaults sometimes) etc. These builds serve 
   single purpose - to discover before release that at least most of 
   pkgsrc builds in DragonFly. I don't use these packages in any of my 
   machines either (except testing machine in which /usr/pkg is removed 
   sometimes three times in a day ;).

Q: But can you take care of such builds?
A: No, I haven't time for that.

Q: But if I provide hardware/scripts/etc for that?
A: No, I haven't time for that.

Q: But what can be done to improve the current state?
A: That's the right question ...

In short we can improve the quality of packages and this is the area where 
everyone can help with bugreports. Just report bugs and problems (with 
fixes, please ;) to the NetBSD GNATS. I can take care of these reports, 
but I don't go to fix software which I don't use unless there is a 
trivial fix. This should help everyone.

And there are binary packages ... we have them now, right? No, these are 
not very usable IMHO for now. These are too irregular for real use, use 
wrong options for me etc.

But this situation can be improved? Yes, someone needs to be found who is 
willing to take care of official _regular_ binary builds. Daily builds 
from stable pkgsrc branch etc. These random builds we have at the moment 
don't help much. There is no clear update path from these etc.

There have been several persons jumped in in the past with "hey, I'll do 
that bulk stuff", but disappearing after that. What we need is person 
taking care of _regular_ bulk builds in long term so that security fixes 
can be applied by users in timely manner etc. So, that users can rely on 
these packages.

If there will be person who is willing to invest his time into maintaining 
builds, we have more to discuss (for example default options for our 
official builds etc), but so far there is no point ...


regards,

-- 
Hasso Tepper