[development] Looking for patch love

Moshe Weitzman weitzman at tejasa.com
Sat Aug 26 18:55:30 UTC 2006

like dries, i'm not 100% certain that these patches point us in the precise 
direction that we want to go. but i do think they move us significantly 
forward, which is about all we ask from a patch. none of us has certainty 
about where this is all headed.

if we don't like working with these flows, we can change the code for the 
following release. this is not a marriage.

>   Disadvantages:
>     - Improved consistency, but not ultimate consistency.  Most things 
> will continue to use the "old style" worfklow (and hook mechanism).

IMO, this is not a real disadvantage. We never had ultimate consistency, and 
never will. We will keep improving "old style" stuff as itches get scratched.

The latest version of the node submit patch lets node modules keep their 
their hook paradigm. the node.module is registering fapi callbacks for them. 
just the function names and params changed slightly.

>     - Not always easier to use or easier to theme.  Not exactly elegant 
> either -- often very verbose.

themeing pertains only to the profile patch. as mentioned elsewhere, i 
expect people to theme using the new user_profile.tpl.php. i agree that 
theme_user_profile is more complicated than before, but thats not really 
important .

>     - Expensive in terms of processing cost.  Performance matters a lot 
> and unless we make Drupal faster, we won't see bigger and bigger Drupal 
> site.

yes, performance matters a lot. it is less relevant for the node submit 
patch, since node submission is a rare event. node submit uses standard 
fapi, so unless we want to open a debate on fapi performance, this is off 
topic. note that most sites do show at least 2 forms every page view (search 
and login), and thats a where i suggest we pour attention if we care about 
fapi performance.

the performance of the user profile patch is more interesting. constructing 
the various fields and categories is virtually unchanged. the rendering is a 
bit more complex in this patch, but we already tested this with the node 
body patch (http://drupal.org/node/74326). this is perfectly analogous, 
except for the fact that it happens far less frequently (and thus is less 
risky). do we really have to benchmark again? this is just a different array 
going through the same drupal_render() engine.

i think i've addressed most of the concerns. i'm willing to keep doing so. 
users can still submit nodes and view profiles. with this patch developers 
can programmatically submit nodes easier, and can view/alter profiles using 
a more standardized method. they can alter profile fields to show/hide using 
#access, and they can recategorize as needed. themers get a new 
user_profile.tpl.php, and well as cleaned up markup in case they want to 
style using only css.

> When discussing, stick to the topic at hand and avoid wandering off to 
> Meta Land to discuss the way we do things in general, or how we should 
> do things.  This is a _technical_ mailing list, and I'm explicitly 
> asking for a _technical_ discussion on this matter.  Don't bother 
> responding unless you spent at least 20 minutes poking at the above 
> patches, or unless you tried an example or two.  Take the time to 
> investigate this, be critic, and share your own opinion rather than 
> repeating what others have said (without having looked at the patches 
> yourself).  There, I've said it. :)

this is a terrific thread starter. i'd like to append it to all my issues.

More information about the development mailing list