[OpenLayers-Dev] moving OL to a scheduled release cycle

[Christopher Schmidt]

On Sat, Jun 09, 2007 at 09:13:54PM +1000, Cameron Shorter wrote:
> I'm tentatively supportive for a fixed release cycle.
> A fixed schedule will make planning to use Mapbuilder easier.
> 
> I'm tentative because:
> 1. There is a reasonable overhead associated with a release cycle, 
> (managing issues, testing etc). I'm guessing that Chris is offering to 
> provide labour, so in that case this is not an issue.

I think it's actually less work if we make a harder cut-off for
features. Right now, one of the biggest things that gets decided is
'when is it done' -- at least for me. This answers that question. Our
actual release process is relatively simple: just follow a couple steps
on a command line, and upload the new tarballs to the server. With
smaller releases, testing is easier because you're breaking fewer
things. etc. etc.


> 2. What happens if we go through the release cycle, and find a 
> show-stopper bug just before we are due to release? Do we release with 
> the bug or hold off. I'd suggest we should hold off.

Absolutely. No regressions or blockers should be going out in releases,
even if we're behind on schedule. Sometimes things slip. However, with a
hard feature freeze 6 weeks in and a target release date of two weeks
later, that should actually give us time to do testing of things --
heck, we could even make those testing periods testing binges, where
developers try to write more automated tests. (It's obvious that our
current testing is not complete enough, because *no* testing is complete
enough, but it's actually not that bad.)

It's good to hear tha tyou're positive on this. 

Regards,
-- 
Christopher Schmidt
MetaCarta