Re: [gentoo-dev] State of Developement

[George Shapovalov]

Hi Grant, Jeremiah

On Monday 26 August 2002 12:28, Grant Goodyear wrote:
> > And they have to go through a specific set of privileged people.
>
> True, although to the best of my knowledge all distributions have
> a similar system.  I don't think it will ever make sense to
> provide CVS write access to all users; it's simply too easy to
> mess things up.  Goodness knows that our developers regularly
> manage to render Gentoo uninstallable, and they're trying very
> hard not to do that.  Our system isn't perfect, but I'm not sure
> what would be better.
Jeremiah: yup, above is quite to the point. I'd second all that Grant has 
said. However there is some hope for more general user involvment. Please 
read below...

> > Q: Can I help, can I help
> > A: Sure, you can take out the trash and scrub the toilets.
>
> Well, right now that's pretty much what the developers are doing,
> too.  Our biggest need for help right now is good bug reports and
> help with existing bugs.  The only major "development" that is
> ocurring right now is portage and some people starting to work on
> a sensible installer.  Everything else is really bug fixing of
> one sort or another.
This is quite true, core developers spend large portion of their time "taking 
out the trash". However I am afraid that we are pretty stuck in this mode the 
way things are now. The problem is: these bugs are mostly not ours, but 
rahter due to ignorance/weirdness/etc of upstream stuff, I mean here original 
authors of the packages. It would be a life in heaven if everything installed 
via "./configure && make && make install". All packages would compile on any 
arch and glibc/gcc version and no weir dependencies were lost in docs and 
nothing would conflict... Well, you get the idea. Unfortunately this is our 
imperfect word where everything tends to break and noticeable portion of bugs 
require getting in touch with package authors or figuring out just what 
build/configuration changes were implemented in this new revision?
   Things will get better when we release 1.4, however I am afraid not by 
much.

The point I am getting at here is that we should and can allow our users to 
take care of a large portion of non-crytical packages. What we need is:
1. Multiple stability levels or KEYWORDS
2. Streamlined ebuild processing - that automates submission of new 
ebuilds/versions and assigns them some kind of "new" or "user-test" 
keyword/level
3.Feedback system that collects user voices and moves ebuilds to corresponding 
categories increasing or decreasing their stability rating.
4. Core group oversees and takes care of the core/crytical stuff and has 
*much* more time to work onportage/sysinit/other_more_distro_specifc_stuff
Well, that would be one possible point of view on this :).

Please take a look at bug#1523 for much more details (though bear in mind that 
some of that stuff is outdated by now).

> > Among all the distributions I am familiar with, Gentoo is,
> > in my opinion, the best as far as placing trust in it's users.
> > But Gentoo is also, in my opinion, far from what I imagine as ideal.
>
> Fair enough.  If you have additional ideas on how we can trust our
> users better, please do post on bugzilla.
That's what I did over half a year ago. Since then I got involved in gentoo 
more than I imagined :).

So, the situation can (and I think should) be changed to give even more 
freedom to our users (and take some stress off core group simultaneously ;)). 
However as any change this takes time, especially with a large evolving 
system which gentoo became. At present we are near completion of step 1 in 
that partial list above (I mean here the KEYWORDS that we were so busy 
adding. As things settle down we will switch to this new masking mechanism 
and get even some more functionality into. At least this was the plan 
according to discussions/announcements over past few month). Eventually I 
hope we will get some more issues resolved and will be offering:
1. multitude of profiles (not just arch/gcc specific but tailored to certain 
tasks: such as server vs desktop, etc.)
2. partial [portage tree] database checkout (according to profile or stability 
setting)
3. streamlined/user_managed ebuild submissions (for non-crytical stuff)
4. anything else?

And it all takes quite some work, so any help is appreciated; but then any 
exciting project has a lot of routine associated with it.

George