On Sun, Mar 02, 2008 at 02:43:56PM +0000, Jeremy Henty wrote:
On Sun, Mar 02, 2008 at 11:31:16AM -0300, Jorge Arellano Cid wrote:
On Sun, Mar 02, 2008 at 02:15:39PM +0000, Jeremy Henty wrote:
Dillo has been neglecting its valgrind cleanliness recently and it turns out the culprit is dw::Table, which resizes its internal vectors without initialising any new values that appear.
AFAIU, this patch covers the symptoms of a BUG that current CVS fixes. Please test current CVS against valgrind.
No change. 121 "Conditional jump..." errors, reduced to 2 by my patch.
To explain, these errors appear because the code updates the entries of the cumHeight and colWidths vectors with wrapper methods setCumHeight and setColWidth. These wrappers begin by comparing the new value with old one and branching on the result, so you get the "Conditional jump..." errors whenever the old value was not initialized. That means these errors have nothing to do with any of the Table algorithms, they're entirely due to the way the Table code manages the cumHeight and colWidths vectors. Jeremy Henty