[development] One core, many distributions

Eric Scouten es-lists at ericscouten.com
Thu Nov 24 21:11:56 UTC 2005

On 23 Nov 2005, at 14:11, Bèr Kessels wrote:

> Op woensdag 23 november 2005 17:15, schreef Dries Buytaert:
>> Clearly, there is a tension between breaking backward compatibility
>> and not breaking backward compatibility.  Unfortunately, there is no
>> "winner" because the costs can't be quantified.  Not the absolute
>> costs.  Not the relative costs.  I'm in the camp that, we are best of
>> breaking backward compatibility when necessary; it buys us
>> maintainability and flexibility, which, in turn, makes for a longer
>> product lifecycle.
> Lets see this is as a test case. But let us measure the "costs" in  
> amounts of
> developers leqving Drupal, or even sticking with 4.6 (and thus not  
> upgrading
> their contribs, or thus not contributing back to bugs and so in  
> 4.7). And let
> us hope that the amount of new developers that are attracted by the  
> easier
> development will at least break us even.

Ber, very nicely stated. I think that's an entirely reasonable criteria.

Unfortunately, the 4.7 cycle has placed me in the drop-out camp.  
Somewhere around August or September, I realized that each CVS pull  
would cost me several hours of development time (downloading from  
CVS; merging into my local Perforce repository; figuring out what  
APIs had changed out from under me -- I have quite a bit of custom  
code; and dealing with many white-screens-of-death). Many of those  
multi-hour sessions were prompted by the false hope that what had  
been broken by the previous CVS pull would be fixed this time around.

After my "day job" (I hate that term, but can't think of anything  
better) and family life, I have maybe 10 to 20 hours per week to  
spend on my web sites. Far too much of that time was going into this  
debugging effort, so I've now backtracked to 4.6.x plus a few  
specific upgrades that are useful to me (notably freetagging).

No value judgement here, just a data point. I'd love to be able to  
contribute bug fixes and some of the custom work I'm doing, but the  
cost of keeping up has become too high. I'll probably wait until a  
month or two after 4.7 is released before considering an upgrade again.


More information about the development mailing list