On December 27, 2002 01:29 pm, Dan Kegel wrote:
I think this patch touches every single TRACE statement that prints out a RECT. Many different formats are used... might be nice to standardize them sometime. (See my related post about maybe using macros for that.)
Now that you've identified all these places, it seems to me the perfect oportunity to standardise them. At the very least, we should convert them all to a common format (say (x0,y0)-(x1,y1), or anything you fancy) so we can more easily grep/sed for them later. But I'd certainly prefer the debugrect() approach, provided Alexandre accepts it.
Dimitrie O. Paun wrote:
On December 27, 2002 01:29 pm, Dan Kegel wrote:
I think this patch touches every single TRACE statement that prints out a RECT. Many different formats are used... might be nice to standardize them sometime. ...
Now that you've identified all these places, it seems to me the perfect oportunity to standardise them. At the very least, we should convert them all to a common format (say (x0,y0)-(x1,y1), or anything you fancy) so we can more easily grep/sed for them later. But I'd certainly prefer the debugrect() approach, provided Alexandre accepts it.
Note that some of these print out the corners, but others output one corner plus width and height. Will anyond mind terribly if we always just print out the corners?
I think I'll start off by switching to the common format you suggest, and using just one macro to deal with arguments. e.g. RECT stuff[20]; ... TRACE("the 6th rect is (%ld,%ld)-(%ld,%ld)\n", RECT_LTRB(stuff+5));
That will make it trivial to switch to the debugrect() approach later, which as someone pointed out involves a bit more infrastructure work (because of the buffer it has to allocate). All we'll need is find . -name *.c | xargs perl -p -i.bak -e 's/(%ld,%ld)-(%ld,%ld)/%s/g' -e 's/RECT_LTRB/debugrect/g'
Ok, off to do a little patch munging... I can resubmit the patch this weekend if I'm lucky. - Dan
-
Dan Kegel -
Dimitrie O. Paun