[PATCH] KIO::SlaveBase and the event loop
nf2 at scheinwelt.at
Mon Jul 21 14:25:55 BST 2008
David Faure wrote:
> On Wednesday 16 July 2008, Roland Harnau wrote:
>> 2008/7/16, Ingo Klöcker <kloecker at kde.org>:
>>> The above "proof" is totally irrelevant as well. When I asked for actual
>>> numbers then I expected actual numbers measured by profiling actual
>>> applications (you mentioned Konqueror) e.g. via valgrind/callgrind.
>>> Unless you can proof that starting lots of IO slaves in Konqueror is a
>>> real problem your concerns are insubstantial.
>> Some logic may help: If starting 20 http slaves is a "problem"
>> because it takes ~2.6 seconds, than it is especially a "problem" if
>> Konqueror starts 20 http slaves during webpage loading, since
>> KIO::Slave::createSlave is the way used by KIO (and hence by
>> Konqueror) to create new IO slaves. And Konqueror/KIO launches in fact
>> over 20 http slaves at some sites. The conclusion is simply an
>> application of modus ponens.
> The planned fix for this is HTTP pipelining support.
Are there any considerations regarding moving from KIO to something like
libsoup? While it seems to make sense for VFS to run backends
out-of-process (like KIO, GVFS and FUSE do it), perhaps that's not
necessary for web-browsers/http. From reading about libsoup i got the
impression that it's becoming a full-fledged http-client library.
Perhaps another area where code could be shared.
More information about the kde-core-devel