[Webkit-unassigned] [Bug 27436] gobject bindings need access to keyCode on KeyboardEvents!

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=27436





--- Comment #19 from Luke Kenneth Casson Leighton <lkcl at lkcl.net>  2009-08-03 16:32:31 PDT ---
(In reply to comment #14)
> I think you should try to enlist one of the longer-term WebKitGtk contributers
> to help you with these patches. 

 *sigh*.  sadly, they've all said "we'd love to, but we don't have time". 
please don't spend time reading any of the 300 comments in #16401, it won't
help :)

> You're hitting a lot of road-blocks at once. 

 well, that's better than hitting one and only one, for eight months straight.

> Much of that is likely due to uploading 15 "first time" patches at once.  So
> you end up hitting similar combinations of starter-mistakes with each.

 yep.  caught lots of them.  cancelled review on ones where i remember.

>  In this
> case, Mark is asking for a re-architecture of how we do autogeneration.  That's
> a little much to expect of a new contributer, but certainly possible. 

 *sssss*... i'm not the person to ask to do that.  not on something that's at
the core of the project.

 and, also, i think it would be a good idea to get the gobject bindings in,
first, _before_ making any significant changes.

 in that way, all the bindings can be considered - and taken into consideration
- all at once.

 if one of them is left out, then... 

> In this case, it may be too much to ask, because our auto-gen system is a pile
> of ugly perl which several of us have wanted to re-write for years.  

 i did as best i could a hybrid cut/paste job.

> Eventually
> one of us will get around to re-writting it in c++ or python.
> 
> I think it would be a relatively easy change to add some sort of
> ALLOW_ARGUMENT_OVERLOADS or similar define which encapsulated !JAVASCRIPT and
> GOBJECT.  That wouldn't involve re-writing the world, but it also wouldn't be
> poluting our IDL files more.

 mmm.... can't give an opinion - bit tired, and also i don't quite follow
 [need context].

 i can say however that the mozilla approach is to simply add extra IDL
attributes, and they have a "no" prefix to indicate !.  e.g. [noscript] means
"exclude from both javascript and xpcom when generating bindings" which they
freely admit is a broken idea.

 as you can see from this:
   https://bug459452.bugzilla.mozilla.org/attachment.cgi?id=342670

 this is an attempt to fix the issue by adding an [optional_arg] IDL attribute,
and then the number of arguments is inserted as an additional argument into the
c++ code.

 i don't like it.

 and, as mentioned here:
   https://bugs.webkit.org/show_bug.cgi?id=27435#c29
 i believe that this whole issue is moot, i.e. from a "practical"
 software engineering perspective, a whole boat-load of complexity
 could end up getting added for very little extra benefit / ROI.

 evidence to the contrary, i _do_ subscribe to the "good enough" software
engineering principle - it's just that with webkit-gobject and pyjamas-desktop
the bar's a bit damn high.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.