I'm not sure if this is a good idea.
I don't think it's safe to assume that stderr will be lost if the program was started from a .desktop file.
If default debug output slows anything down noticeably, that should probably be fixed.
Vincent Povirk
On Fri, Apr 3, 2009 at 2:40 PM, Austin English austinenglish@gmail.com wrote:
Users aren't running in a terminal, so debug output only slows things down.
-- -Austin
On Fri, Apr 3, 2009 at 3:19 PM, Vincent Povirk madewokherd+8cd9@gmail.com wrote:
I'm not sure if this is a good idea.
I don't think it's safe to assume that stderr will be lost if the program was started from a .desktop file.
The majority of users are double clicking .desktop files to start their applications. Where do you see stderr going to in this case?
If default debug output slows anything down noticeably, that should probably be fixed.
Several games experience this problem, but it's difficult to use the quiet fixme approach, since they have variable information. Converting the fixme's to trace's/warn's may be a viable option.
On Fri, Apr 3, 2009 at 3:29 PM, Austin English austinenglish@gmail.com wrote:
Several games experience this problem, but it's difficult to use the quiet fixme approach, since they have variable information. Converting the fixme's to trace's/warn's may be a viable option.
What if we abandoned "print FIXME the first time, print nothing subsequent times" in favor of "print FIXME the first time, print WARN|TRACE subsequent times"?
I've personally lost some useful information in a debug log because of the "FIXME once" idiom. I was explicitly asking for the gdiplus channel and expected to see all the calls.
A FIXME_ONCE macro could be written fairly easily that does this efficiently (i.e. without writing the same code twice).
Vincent Povirk
On Fri, Apr 3, 2009 at 3:38 PM, Vincent Povirk madewokherd+8cd9@gmail.com wrote:
On Fri, Apr 3, 2009 at 3:29 PM, Austin English austinenglish@gmail.com wrote:
Several games experience this problem, but it's difficult to use the quiet fixme approach, since they have variable information. Converting the fixme's to trace's/warn's may be a viable option.
What if we abandoned "print FIXME the first time, print nothing subsequent times" in favor of "print FIXME the first time, print WARN|TRACE subsequent times"?
I've personally lost some useful information in a debug log because of the "FIXME once" idiom. I was explicitly asking for the gdiplus channel and expected to see all the calls.
A FIXME_ONCE macro could be written fairly easily that does this efficiently (i.e. without writing the same code twice).
I think that would be very useful, and would love to see that.
However, that's a separate issue from disabling output for .desktop files. I don't see many (any?) use cases where having that output when launched from a gui is useful.