[development] Switching themes in modules

[Bèr Kessels]

Op donderdag 22 juni 2006 23:59, schreef Dries Buytaert:
> Is that why you forked core's upload.module ?

It was not a fork. It was a complete rewrite ground up. But I started off with 
upload.module. A diff will show you that only five tiny chunks remain intact 
and that all the other stuff was changed. Not only large clean ups (really: 
upload.module has "functions" that are like 200 lines big!!), but also 
removing lots of stuff from upload into file.inc where it belongs. Not only a 
complete overhaul of the file.inc and the upload.module, but also introducing 
a FAR more modular and flexible file system!

Besides, And that is the main reason for chiming back in, I *did* work on 
upload.module. I maintained a patch for months that allowed SIMPLE inline 
file handling. That patch got loads of "heavyweight" positive votes. Yet 
somehow you  (Dries) did not want it. What am I supposed to do then? I am not 
a fan of sending bribes, nor of sending horse heads, when a (badly needed) 
patch is refused. 

That made me decide that a) pulling something out of core to develop on in in 
a "free", pragmatic way (as in: not maintaining broken patches for ages, but 
working on it over/with CVS), and b) rewriting stuff from ground up, is the 
way to go. Not fixing tiny pieces of old code line by line, patch per patch.

I think the amount of cruft and inconsistencies in core show us that the 
iterative way of working does not work well enough. I hoped that pulling 
something out of core, and deconstructing (aka rewriting) it completely did 
work. Because by the time it is stable and finished, it most probably is far 
better then the stuff in core will ever be. In far less time, with far less 
patch-rewritings, etc.

I know people disagree with me. I know a lot of people think "well then: fix 
it". I know that even more people think "sigh, there he goes again". But tell 
me: why the ## do we have so much stuff in core, from which everyone agrees 
that it should not be there, or that it badly needs a rewrite? Yet I hardly 
ever see patches/issues like "Complete clean up of hook_user", or "removed 
archive.module from core", or "complete clean up of the upload.module code", 
get trough, or being made at all? The only answer I can give is: "Drupal is 
too large, for the way it works now". I don't have a ready made solution 
either, but the fact that we have so much cruft in core, so much good patches 
in the queues collecting dust, and the fact that we still don't have a proper 
media handling system shows that this is somehow right!

In any case, this very patch, the reason for this "fork", which was never a 
fork, made me loose a lot of confidence in Drupal and its way of working. 

Bèr