[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-devel
Subject:    Re: low performance of kio
From:       Stephan Kulow <coolo () kde ! org>
Date:       2000-04-20 20:30:35
[Download RAW message or body]

Simon Hausmann wrote:
> 
> On Thu, 20 Apr 2000, Stefan Schimanski wrote:
> 
> > > > Forgive me if this sound naive, but I was under the impression that KIO should be used
> > > > whenever possible, in order to promote said "transparency". If you insist that ppl
> > > > use which ever method is more appropriate, then use fread for local files and nfs or
> > > > ftp for network operations, and skip KIO altogether.
> > > >
> > >
> > > This is kindof what I was thinking but I was too much a newbie with KIO
> > > to mention it ;-) I thought one of the main advantages of KIO was to be
> > > able to use the same chunk of code for both local and remote files >:)
> > > I'm totally ignorant here but perhaps it would be good to add hacks to
> > > KIO that optionally just use normal I/O without slaves for local files
> > > but generate the same signals as other jobs (reading/writing a chunk of
> > > data then generating a signal)? Dunno if this is possible or not. It's
> > > certainly not good to be saying "Don't use KIO for local file access".
> >
> > I agree. This would speed up several apps a lot. Loading huge pictures is quite
> > slow with the current implementation. Another example is konqi. Try to copy a
> > 600 MB cdrom image with it. Be sure to reserve enough time to copy it with
> > around 5 sec per MB :-)
> 
> The point about implementing this is that it would completely kill the
> idea of non-blocking IO. This might not be critical for real local files,
> but it is for transparent file systems like nfs or smbfs. (ever
> experienced a hanging nfs mount?)
> 
> I guess the advice is: If you want non-blocking IO for "anything" (full
> transparency for all supported protocols) then use KIO. If you don't care
> if your GUI app blocks for reading/writing and don't need the
> transparency, then use fread and friends.
> 
> But I'm no KIO hacker, I'm just saying what they told me ;-)
> 
Well, luckily enough I once got code by Olaf Kirch to implement
is_probably_slowly_mounted()
that could be used to put the code in place :)

Greetings, Stephan

-- 
It said Windows 95 or better, so in theory Linux should run it
                                                GeorgeH on /.

[prev in list] [next in list] [prev in thread] [next in thread]