http://starurchin.org/dillo/valgrind/10ca58f617a0693638facdb40bcf319af1bdfc6... has a leak at dw/table.cc:323, and I found a way to trigger it: <table border=2> <tr><td>first<td rowspan=2>tall<td>third <tr><td colspan=2>wide<td>last </table> Jorge, you might remember a little something about tables. Can we simply check children->get() before putting anything there, or do we have to fix things somewhere if part of a span is stolen away?
Hi there, I'm back from a new machine (still tweaking it, but workable!) On Sun, Jun 05, 2011 at 03:09:35AM +0000, corvid wrote:
http://starurchin.org/dillo/valgrind/10ca58f617a0693638facdb40bcf319af1bdfc6... has a leak at dw/table.cc:323, and I found a way to trigger it:
<table border=2> <tr><td>first<td rowspan=2>tall<td>third <tr><td colspan=2>wide<td>last </table>
Jorge, you might remember a little something about tables. Can we simply check children->get() before putting anything there, or do we have to fix things somewhere if part of a span is stolen away?
Hmmm, fixing it from the HTML parser would be harder as AFAIR it doesn't know much more than adding rows and cells. Corner cases as this should be easier to handle in the dw-table as you suggest. It looks like an overlooked case, or maybe just paying the "leak cost" of it. ;-) You should review the algorithm to make sure the stolen cell will not be missed elsewhere. -- Cheers Jorge.-
participants (2)
-
corvid＠lavabit.com -
jcid＠dillo.org