On Thu, Jun 10, 2004 at 15:16:41 +0200, Thierry Vignaud wrote:
Daniel Kasak <dkasak nusconsulting com au> writes:As for your second question, I think that using Glade is the only way to go. Others will disagree with me. But using Glade means that you've got more time for actual coding, and also that modifications are much easier - especially when you're building complicated forms, and *especially* if someone else has to modify your creation.maybe do one save some time when prototyping a GUI, but when you'd to change sg, you'll have to change it everywhere whereas with real code, you can factorize common stuff (compare with HTML+CSS or LaTeX against HTML as found in most web pages) thus you eventually loose time while maintaining your application and it may be error prone (eg: one forget to update a place).
Yes, it might be a problem, sometimes. The key to this is full use of
possibilities glade allows. I especialy mean binding signal handlers and
also custom naming of widgets. That way layout changes don't need to
touch perl code and functionality changes do not need to touch the glade
file (no, I never use the default handler names -- the handler name
should be derived from what it does and not what widget it is bound to).
You can also use realize or show signal for 'initialization' -- so the
initialization code does not have to match names -- and you can even
write perl core to the signal handlers. I use this to overcome the
problem, that user data for signal can't be set from the builder (though
the library supports it). But it was not accepted into the distribution,
as it was considered bad style to have code in glade files.
-------------------------------------------------------------------------------
Jan 'Bulb' Hudec <bulb ucw cz>
Attachment:
signature.asc
Description: Digital signature