Am 06.05.2014 18:43, schrieb Eric Anholt: > walter harms <[email protected]> writes: > >> Am 06.05.2014 00:02, schrieb Keith Packard: >>> This allocates span data for all of the provided arcs and draws the >>> whole set in one call, rather than doing them one at a time. For >>> modern hardware, this is a significant performance improvement. >>> >>> Signed-off-by: Keith Packard <[email protected]> >>> --- >>> mi/mifillarc.c | 98 >>> ++++++++++++++++++++-------------------------------------- >>> 1 file changed, 34 insertions(+), 64 deletions(-) >>> >>> diff --git a/mi/mifillarc.c b/mi/mifillarc.c >>> index 337343d..1695121 100644 >>> --- a/mi/mifillarc.c >>> +++ b/mi/mifillarc.c > >>> + >>> + pts = points = malloc (sizeof (DDXPointRec) * nspans + sizeof(int) * >>> nspans); >>> + if (!points) >>> + return; >>> + wids = widths = (int *) (points + nspans); >> >> >> this looks like a case for calloc(). > > calloc also does the work of zero initialization, which you don't want > in your rendering paths where the data is getting fully initialized > anyway. yep, but it is more clean and it (should) takes care about integer overflows (maybe not a problem here). Later version may not be fully initialized. NTL this is my opinion ... if you think it does not matter leave it.
re, wh _______________________________________________ [email protected]: X.Org development Archives: http://lists.x.org/archives/xorg-devel Info: http://lists.x.org/mailman/listinfo/xorg-devel
