On 10/11/2012 20:35, Jonathan Perkin wrote:
* On 2012-10-11 at 19:17 BST, John Marino wrote:That leads to another tangent: No build farms. In an ideal (fantasy) world, somebody would throw their proposed changed in a build farm and ensure that no regressions (build failures) were suffered on any platform. Pkgsrc BADLY needs something like this. I know it costs money, I'm sure we could scrape together funds for a dedicated machine but it still has to be hosted and controlled with other machines. I don't see any talk about a facility like this, but I think it's essential to keep QA levels up.We (Joyent) have a pkgsrc build farm. It's not public access, but I am more than happy to run a bulk build for a particular change if someone is worried it will break lots. If you put the change up on github then it's even easier. SmartOS only, though, of course :-)
Right, that's kind of you. I was thinking along the lines of what FreeBSD port guys do. I don't know how automated it is, but every proposed change to a port is run through a tinderbox for every architecture they support. If it passes them all, then it gets committed.
What that would look like for pkgsrc (I guess) is one machine (or virtualized machine?) per platform to test. You might two DragonFly machines, 4 NetBSD machines, a couple of Linux machines, smartOS etc. It would be nice to submit the change and have all the machines try to build the package. This type of thing would have caught numerous regressions seen this year, regressions on very important packages.
I know it's a very tall order, but its seems as pretty fundamental to me. Now can SmartOS virtualize all these potential platforms? :)
John