On 11/07/2013 09:18 PM, Rich Freeman wrote: > On Thu, Nov 7, 2013 at 3:08 PM, Denis M. <g...@politeia.in> wrote: >> On 11/07/2013 08:59 PM, Matthew Thode wrote: >>> iirc, we give $200 if infra for developer accounts for a couple of >>> months. If a deal is struck it would likely be more and forever or >>> something. >> I've been running my VM for Ago for 13 months now (started on september >> 2012), where are my >$200? ;-) >> > Can't argue with that. :) > > Seriously, though, I'd love to see these needs better supported. I > think we need to start by defining what the needs actually are (less > redundancy, more consistency, etc). Then we figure out how to best > address them. It could be individuals donating VMs, or it might be > Gentoo buying resources from any number of vendors, or it could be > Gentoo going out and looking for donors. I suspect that if we went > out with something specific in mind we might be able to find a sponsor > - but it is always best to have some idea just what we're going to be > using any donations for (this will be our stage3 builder which cranks > out a new stage3 every 20 minutes and reports build failures to double > as a tinderbox, etc). > > Rich >
Currently Diego's tinderbox does something like that AFAIK. Compiles things and (almost?) automatically submits bugs against the packages with the relevant logs, etc... The initial idea behind my suggestion was that the devs would have the enough system resources to address these bugs (and the ones reported from the users, of course). An example here could be the following: finding/confirming a compilation bug for a package with ~10 USE flags could take tatt quite some compilations depending on the USE flag's combinations (this is actually what arch testers do in order to stabilize/keyword a package). Another example would be, as I mentioned in my previous mails to this thread - a new glibc version comes out and (as you know) quite some packages fail to compile against it. Having the resources, it would be possible to track these packages faster instead of relying on random users/testers to report them to bugs.g.o. And a last one would be testing new KDE/GNOME/whatever-meta-with-huge-number-of-packages. As an AT member myself I could only give examples on how using such system of donating/providing instances would be a benefit. For a comprehensive list of the tasks (for consistency as you said), I'd wait for actual devs to enumerate their needs. I doubt this will go as further as Gentoo actually *buying* resources. The reason is obvious - "things have been going fine till now, why throw monnies for something as 'unnecessary'" (which is why I haven't received a penny for it, hehehe), that's why I came with the donorship-of-instances version. I believe the 'going out looking for donors' part you said is basically what I'm suggesting here, although I believe you meant donors = huge companies providing clusters, and I doubt that'll happen. From my observation, you can get a lot of work done on a simple 2GB-ram-4-cores VirtualBox VM. Not to talk that lots of people nowadays have these resources to spare. That's why getting actual people (and not companies or whatever) to donate their system resources is easier to get/reach. Regards, Denis M.
signature.asc
Description: OpenPGP digital signature
