On Thu, 8 Sep 2005 10:22:43 +0900 Carsten Haitzler (The Rasterman) [EMAIL
PROTECTED] wrote:
On Wed, 7 Sep 2005 22:29:47 +0200 Tilman Sauerbeck [EMAIL PROTECTED]
babbled:
Hi guys,
patch attached, objections? ;)
i have no objections as such... but i am not sure of the usefulness (at
On Thursday, 08 September 2005, at 19:10:45 (+0200),
Tilman Sauerbeck wrote:
True. What about just calling the shutdown stuff via atexit()?
Does it really matter?
Using atexit() could potentially conflict with client programs' use of
that same function.
Michael
--
Michael Jennings (a.k.a.
Well, potentially.
The atexit funcion registers the functions on a LIFO stack, soo it
will be called, no matter if the client program also registers
functions on it.
Altough it may be bad if the client uses the shutdown function before exiting.
But i don't see much of a problem here. IMHO.
On Thu, 8 Sep 2005 19:10:45 +0200 Tilman Sauerbeck [EMAIL PROTECTED]
babbled:
On Thu, 8 Sep 2005 10:22:43 +0900 Carsten Haitzler (The Rasterman)
[EMAIL PROTECTED] wrote:
On Wed, 7 Sep 2005 22:29:47 +0200 Tilman Sauerbeck [EMAIL PROTECTED]
babbled:
Hi guys,
patch attached,
On Friday, 09 September 2005, at 09:56:22 (+0900),
Carsten Haitzler wrote:
well i think i'd rather the shutdown and init calls for explicit init and
shutdown than atexit() hooks myself :)
As long as everything works if you don't use those functions, I guess
that's not so bad. But breaking
On Thursday 08 September 2005 08:56 pm, Carsten Haitzler wrote:
On Thu, 8 Sep 2005 19:10:45 +0200 Tilman Sauerbeck [EMAIL PROTECTED]
On Thu, 8 Sep 2005 10:22:43 +0900 Carsten Haitzler (The Rasterman)
On Wed, 7 Sep 2005 22:29:47 +0200 Tilman Sauerbeck
Hi guys,
patch attached,
On Thursday, 08 September 2005, at 21:47:14 (-0400),
Mike Frysinger wrote:
you could make the whole thing transparent by declaring them as constructor
and destructor functions
int imlib_init (void) __attribute__ ((constructor));
int imlib_init (void) { do your init thang }
int
On Thursday 08 September 2005 10:21 pm, Michael Jennings wrote:
On Thursday, 08 September 2005, at 21:47:14 (-0400),
Mike Frysinger wrote:
you could make the whole thing transparent by declaring them as
constructor and destructor functions
int imlib_init (void) __attribute__
On Thursday, 08 September 2005, at 22:40:05 (-0400),
Mike Frysinger wrote:
it's pretty neat though and not like we cant just macro it away so
that the cool hosts (like linux) benefit from it
And what fallback do you propose for other platforms? And how would
programs that utilize this feature
On Thursday 08 September 2005 10:47 pm, Michael Jennings wrote:
On Thursday, 08 September 2005, at 22:40:05 (-0400),
Mike Frysinger wrote:
it's pretty neat though and not like we cant just macro it away so
that the cool hosts (like linux) benefit from it
And what fallback do you propose
On Thursday, 08 September 2005, at 23:04:21 (-0400),
Mike Frysinger wrote:
not everyone cares about portability,
I know, and I've been fighting that fight for 10 years now.
so refusing a simple but neat feature
Sorry, but it's really not neat at all. It's just one more way to
cause things
Hi guys,
patch attached, objections? ;)
Regards,
Tilman
--
learn to quote: http://www.netmeister.org/news/learn2quote.html
imlib_init_shutdown.diff
Description: Binary data
On Wednesday, 07 September 2005, at 22:29:47 (+0200),
Tilman Sauerbeck wrote:
Hi guys,
patch attached, objections? ;)
A bit late for API changes, don't you think?
Michael
--
Michael Jennings (a.k.a. KainX) http://www.kainx.org/ [EMAIL PROTECTED]
n + 1, Inc., http://www.nplus1.net/
On Wed, 7 Sep 2005 22:29:47 +0200 Tilman Sauerbeck [EMAIL PROTECTED]
babbled:
Hi guys,
patch attached, objections? ;)
i have no objections as such... but i am not sure of the usefulness (at this
stage) as imlib2 has been around for so long post 1.0 :)
--
- Codito, ergo sum - I
14 matches
Mail list logo
