Re: Good news, testers required

On Thursday 04 October 2007 11:14:22 Andras Mantia wrote:
> Hi,
>
>  I finally found the cause of the 3.5.7 slowness. It was introduced by
> some safety checks to avoid crashes in the parser. Unfortunately this
> checks made the parser VERY slow, so no wonder people seeing several
> seconds delays with some bigger documents.
>  I changed the checks, while still keeping them. The result is
> impressive, my big test document (700KB, 12000 lines) parses in about
> 350ms compared to the 3.5.7 version where the same operation took
> 11-14seconds(!). Switching away from this document to an empty one took
> no less than 48-50seconds, while now it is around 60ms.
>  Also the checks were not completely correct, so 3.5.7 could still
> crash. Now 3.5.8 not only that it should be stable, but it should be
> also almost as fast as the previous 3.5.x versions, prior to 3.5.7.
>  3.5.8 will be tagged on Sunday, but I ask those who are used to compile
> from source to download the latest version from the Subversion
> repository[1] and test it. 

ok, did so - works ok here.
w.r. to speed, I cannot say very much, as 3.5.7 worked for me too,
probably because I usually use to work on distinct parts of my project(s), 
having only few (~5) relative small (mostly .php) files open.
however, thanks alot for the fix/improvements, Andras.

werner

_______________________________________________
Quanta mailing list
[email protected]
https://mail.kde.org/mailman/listinfo/quanta

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQBHBSLaKpc4k0KcOP4RAiSkAKCxXeLVEX3pMGxYeMendb74vyz7LACeLCui
n4rQtCROXmt8QHeBsb3JPSg=
=HYvA
-----END PGP SIGNATURE-----