On Mar 26, 2007, at 8:05 AM, [EMAIL PROTECTED] wrote:
> > Hi: > > I just read the post here > http://programming.reddit.com/info/14ays/comments/c14enf > > They have some debate about TurboGears and other framework. > > There's an critique said: > > CherryPy 3 was released almost two months ago and kicks ass. I want to > use it. SQLAlchemy is way more powerful than SQLObject. I want to use > it. Genshi is like Kid but better, I want to use it. Given this state > of affairs, do you really think choosing SQLObject, Kid and CP2 is > "immediately obvious to anyone who has actually looked at the > subject"? Because those certainly aren't the tools I would recommend > as a knowledgeable developer. > > > I think it's a solid comment, and we should do something to fit > people's expectation. Or we'll lose most of new people. I think we all agree on this. The port to CherryPy 3 is almost done in the cp3 branch, just needs the tests adapted as the current ones are too closely tied with CP2's internals (I'll give a shot at this in a few days) and widgets ported from 1.0 as the ones in trunk give the wrong impression if being 100% usable with Genshi which is not the case. Once this is done we can merge that branch with the trunk and release a 1.1 alpha. This reminds me of something important: Any patches that are merged from 1.0 to trunk should also go into the cp3 branch until it's merged. Hopefully this situation will end soon but please try to keep an eye on it now so we don't miss anything. > I think we should release 1.0.2 ASAP (at least give it a date) and > toward next step, ex: test CP3 branch and make some docs about genshi/ > sqlalchemy, There are a couple of pending tickets and patches which need to be commented on, applied or discarded. New features should not be committed IMO unless they are 99% risk free and obviously missing from what the API suggests should be possible (I know, this can be very subjective.... that's what this ML is for... :) It would also be a good idea IMO to start enforcing a "one change, one test" policy which will make the committers job much easier and raise the overall confidence when applying new patches. I can't spend much time on this ATM although I hope I'll have more free time in mid-April. That shouldn't be a problem as I'm not the only committer, am I? ;) Really, I strongly believe every committer who has been using TG for a while and has some idea of how things work internally is perfectly capable of applying patches. If there's any doubt... that's what this ML is for... :) If something goes wrong, "svn merge" is there to the rescue. Anyway, I see no reason why we should "serialize" 1.0.2 and 1.1. Work can be done in parallel in both branches by whoever is motivated to work on them. Alberto --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "TurboGears Trunk" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/turbogears-trunk?hl=en -~----------~----~----~----~------~----~------~--~---
