Re: Refactoring

[Rob Taylor <rob taylor codethink co uk>]

Philip Van Hoof wrote:
> On Thu, 2007-11-22 at 18:09 +0000, Rob Taylor wrote:
>> Philip Van Hoof wrote:
>  
>>>    I know a lot of glib fans will disagree with the necessity for this
>>>    (hey Rob!), nonetheless is this actually needed indeed.
>> So, does that also involve removing GObject support? I presume so as the
>> only reason you could want to do this is to build without a glib
>> dependency, right? If so, I'll let Mark know he doesn't need to bother
>> continuing with the pygobject based bindings..
> 
> I don't think I will ever stop using GObject nor will I stop depending
> on glib. It's just the GMainLoop part that I no longer want to consume
> unless the toolkit uses it.
> 
> In fact am I planning to rewrite some of the abstract types and
> interfaces using Vala. Unless Jurg is planning to implement a backend
> that doesn't use GObject ... 
> 
> Differently put, I want the libtinymailui library to implement an
> abstraction for me, and consume the abstraction. While for
> libtinymailui-gtk that implementation will just use GMainLoop, of
> course.
> 
> Using a GObject instance in for example Qt, is not problematic. Using
> GMainLoop with recent versions is probably not problematic either. But
> if it's more ideal for both old and recent versions of Qt to use their
> mainloop ... then I would love tinymail to do that in stead.
> 
> 

Ok, I guess that makes sense! If I were doing the work (which I'm not
;)) I'd personally leave doing an abstraction until there's a known
couple of use cases, otherwise you might end up doing the work twice!

Of course, a general (liberally licensed) mainloop abstraction would be
of general benefit, I think.

Thanks
Rob

-- 
Rob Taylor, Codethink Ltd. -  http://codethink.co.uk