Open Souce Software is about contributing code to the original sources. Ideally that code would be integrated by the developer quickly. When that is not done good usually a fork happens or a semi-fork(plugin, code-patch ..) happens.
For Gentoo some semi-forks have already happened where the process of integrating contributed code is not good/efficient/quick enough, there we need to improve our processes:
– ebuilds in overlays, we have a lot of overlays today
This increae is clearly due to the fact that we are not able to recruit developers in a reasonable timeframe. Thus they are forced to develop externally and force all of our users to add an overlay for popular applications.
The big step that we need to do here is to offer more external Gentoo developers to get @gentoo.org adresses and the official “Gentoo developer” status. The main problems holding that up as I observe are a) externals being slow on quizzes and b) devrel being slow in checking quizzes. Maybe devrel should put out another call for new recruiters when they know they are short of those.
– gentoo-wiki has for many tasks like my Macbook or Beryl superceeded gentoo.org/doc/en
Here I don’t see the downside of having it external. Unlike for ebuilds it is good to have two sources for information.
– the external pkgcore and paludis projects to superceed portage
I think this is a big downside because portage devs have one more reason now to say “no” for contributions – they can just point to the forks. But that does not help much as the are not 100% back-and-forth switch compatible and not a full replacement either. It just confuses our users to have those and does not help the general problem.
For example this bug of me is stalled:
138792 dobin etc. should automatically die on failure
The big step that portage development needs to take is to focus on pkgcore or paludis and drop portage development in favour of that. I think that step will increase the popularity of Gentoo again. No one wants to use a distribution with a really slow package manager.
>This increae is clearly due to the fact that we are not able to recruit developers in a reasonable timeframe.
I’ve been thinking about helping out for the last 2 years now. But with at all the flamewars on gentoo-def/irc/everywhere, having some people react like 5 year olds on bugzilla, etc. , well, let’s just say I’m more than happy with my local overlay. I just don’t want to be involved in such a hotile environment… The day it seems fun again is the day I’ll change my mind…
>No one wants to use a distribution with a really slow package manager.
Frankly, I don’t care if portage does it’s thing in a flash or not. I care if portage does a good job compared to other distributions. And to be honest, compared to the rpm and deb distros I know about, portage beats them hands down. Who cares if it takes a couple of seconds/minutes? It seems you have forgotten the joys of working with gentoo. Switch back to debian form a couple of weeks, then come back.
This post has just about as many good ideas as sunrice did.
Dan: And your comment has about as much constructive criticism as a 5-year-old can muster. This isn’t atypical of Sunrise’s long-time detractors however.
From anonymous coward:
Other distros’ package managers have progressed, closing the gap between them and Gentoo. And going looking for overlays to get what the user wants isn’t so much different for trying to find the right rpm or deb repository for another distro; it’s very annoying and certainly puts off new users.
I miss the days when all I had to do to get what I wanted was to search through the tree and start emerging and configuring.
I don’t blame you at all for not wanting to deal with the very badly behaved gentoo devs; I hope it will become fun again, welcoming in people who want to help keep the tree healthy.
We are slow at hiring developers because there is a big difference between writing a few ebuilds and then writing ebuilds to Gentoo’s standards. The Gentoo environment is no longer dead simple like it used to be and that is a big problem. It requires training, people knowing about cache regen issues, portage, mirrors, eclasses, testing, etc…
If writing ebuilds was all that devs did we would have 100’s more developers. I think our quizzes should focus more on the other bits of being a dev and that is why recruiters are working on it.
Gentoo-Wiki has always been a good resource; I don’t think it replaces gentoo/doc/en, but as you stated you can’t really have too much documentation.
As for the portage comments, you need to talk to gentoo as a whole. ‘The Portage Team’ is just a few people who are interested in working on portage. I’ve tried working on pkgcore and I haven’t done anything with it. Portage is the only thing that interests me right now. So convince gentoo to switch and portage can fork off on it’s own. However this is already happening, (it’s why we have EAPI/PMS and so forth). If you think we can magically just switch hundreds of thousands of users to a different package manager…well I think thats a very niave view.