On Sat, Jan 2, 2010 at 9:01 AM, Valentin Villenave
<address@hidden> wrote:
On Fri, Jan 1, 2010 at 6:49 PM, David Kastrup <address@hidden> wrote:
Don't load this as a picture into firefox on X11. It is something like
10000x10000 pixels, and apparently firefox allocates a full-size
display-depth pixmap for that, making the server swallow more than 1.2G
of memory.
I managed to kill the firefox process by now. It was not pretty.
I had a look without any problem (well, perhaps having a 64-bit CPU
with lots of memory helps in this regard :-).
Carl, Marc, how would that affect the regtests-checking process? When
applying this patch all outputted PNGs will look different than
before, so isn't there a risk that some actual regressions go
unnoticed?
The regression comparison is not generated from the .PNG files, but
rather from the Stencil bounding boxes at the end of the formatting
process. I suspect that the change does not affect the clef's
bounding box.