|
From: Jan E. <je...@in...> - 2014-05-01 22:55:33
|
On Friday 2014-05-02 00:15, Markus Hoenicka wrote: >Am 2014-05-01 18:50, schrieb Jan Engelhardt: >> I have identified another bug and prepared a patch for >> dbd_mysql/dbd_msql. The purpose of result->currowidx was not totally >> clear-cut, so I am posting this patch for potential comments. > >[...] was designed to avoid an expensive call to mysql_data_seek() >in *every single* iteration of stepping through a result set. There >are data in the mailing list archives which show that MySQL is >slowed down abysmally [...]. It would be great if we could find a >way to synchronize result->currowidx with the DBD row pointer in a >cheaper way. That is the answer I was looking for. I can cook something up now :) |