[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Gnash-dev] Re: DEBUG_STACK
From: |
strk |
Subject: |
Re: [Gnash-dev] Re: DEBUG_STACK |
Date: |
Tue, 25 Jul 2006 16:03:09 +0200 |
On Tue, Jul 25, 2006 at 07:51:26AM -0600, Rob Savoye wrote:
> Sandro Santilli wrote:
> > On Tue, Jul 25, 2006 at 05:33:20PM +0400, Vitaly Alexeev wrote:
>
> > This might be possible the whole problem you detected with
> > general slowdown, and has been introduced since Rob
> > committed the new code dropping the IF_VERBOSE_ACTION macro
> > and friends. I think I've sent a mail on gnash-dev in the past
> > about this.
>
> What happened with change is we switched from a printf() style debug
> log, to one that supports iostreams. If debugging isn't enabled, none of
> that code should get executed, but yes, using the debug log slows down
> the performance. Without -v, it should run at the same speed as it used to.
Ok, I took a close look. Actually, in the stack dump case we're
filling a stringstream and *then* passing it to log_action.
This means the dumper are *always* called.
Still, I think that even directly calling
log_action("%s", function_returning_string());
will always execute 'function_returning_string()', even w/out -v specified.
Am I wrong ? Would:
dbglogile << function_returning_string();
do any better ? Or is a C-style macro our only chance ?
I liked the C-style macro even to be able to completely drop
debugging at compile time.
--strk;