5.0 and or 4.8 (was Re: [development] Drupal x.x.0 freeze date)

[Bèr Kessels]

Op maandag 1 mei 2006 20:08, schreef Angie Byron:
> To be fair, half the reason FAPI took so long to troubleshoot/bug-fix is
> because it was completely new, never-before done and basically a complete
> rewriting of how Drupal does forms (which are used in almost every file ;))
> from the ground up. There was no way to predict all the possible scenarios
> that would need to be accounted for until much, much later.

Yes. Some code exists and a lot does not. CCK might be cool and working. But 
it is far from ready to distribute *instead* of our default nodes. What about 
peoples data, for example. No. The changes I outline are serious, big 
monsters. 
Stuff, that we (a natural Developers problem?) tend to think "a couple of 
hours coding", "a simple update routine". Done in hours. 

No. We thought fAPI would take a few months. It took seven. I think that even 
the most realistic amoungst us would have laughed when someone said "this 
will take over half a year to get right". And we are still not there. Don't 
forget that either!

> Most of the changes you outline either already have existing code in
> contrib or wherever to draw from, or are re-tooling/enhancement type of
> things (like FAPI improvements or X as nodes). I don't see a lot of "start
> completely over from scratch" stuff here, so I think "years" may be a tad
> bit unrealistic.

fAPI, when it hit core, was in development for a long while too. So you can 
safely add a few more months of extra development to my outlined seven. 

Also. We should not forget that besides the fAPI there were no other big 
projects going on. There were some AJAX improvements. And we had a huge 
amount of small issues in queue. But nothing of the same magnitude. 

Can you imagine the mess, if two such large projects are intervening 
eachother? Breaking eatchothers patches all the time? If two such projects 
drain all resources from our developersbase?

I can. And hence I remain at my point of view, that we:

* either need some very strong gatekeeper "thing" in place to manage core, and 
save it from breaking beyond usable ways. 
* Or can be assured that the first four to five months will get core in such a 
state, that it will take three-four times that time (a year, or more) to 
bring it back to a usable state. 

Call me a pessimist, if you want. But I manage to keep my deadlines within a 
10% range because I plan "pessimistic". I call that realistic. People laughed 
at me, when back in november I said that "a release might not even be there 
at the end of february". So. Yes. Unless we plan more carefully this time, we 
will fall in the same traps again. And with the grown developers and 
userbase, talking about "years" is IMO not so unrealistic.


Bèr