Fwd: Removing disabling of exceptions/properties/signal-handlers

[Jonathon Jongsma <jonathon quotidian org>]

apparently I accidentally only sent this reply to daniel.  forwarding
to the list.


---------- Forwarded message ----------
From: Jonathon Jongsma <jonathon quotidian org>
Date: Wed, Dec 30, 2009 at 9:47 AM
Subject: Re: Removing disabling of exceptions/properties/signal-handlers
To: Daniel Elstner <daniel kitta googlemail com>


On Wed, 2009-12-30 at 15:58 +0100, Daniel Elstner wrote:
> Am Mittwoch, den 30.12.2009, 09:51 +0100 schrieb Murray Cumming:
>
> > I doubt that we can remove the no-exceptions build because disabling
> > exceptions is still very popular, mostly out of habit, and is often
> > enforced by awkward hand-built build systems.
>
> But it is the single most important bit, by far.  If you have to cater
> for both cases, with and without exceptions, you may as well not offer
> exceptions in the first place.  It leads to a fundamentally different
> style of programming.
>
> In other words, a gtkmm without exceptions is not the same toolkit
> anymore.
>
> Also note that we already rely on RTTI.

I agree with Daniel.  The exceptions switch is the one I really want to
get rid of.  It makes wrapping libraries hell: if any functions need to
be hand-wrapped, you need to #ifdef tons of stuff and it becomes a giant
pain.  Also, as daniel said, it requires a totally different style of
programming, so making an application work with both styles of gtkmm API
requires a ton of extra work, so nobody ever actually does it.  Which
means that almost no gtkmm applications will actually run on maemo
without a lot of work to port them.  Because of this, there are hardly
any gtkmm applications available for maemo, which means that all of the
work done to slim down gtkmm on maemo seems a bit pointless since almost
nobody actually uses it.

--
Jonathon Jongsma <jonathon quotidian org>




-- 
jonner