On Sun, Sep 19, 2004 at 02:29:16PM -0300, Trevor Smith wrote:
> On September 18, 2004 12:42 pm, Ow Mun Heng wrote:
> > Wel.. could it be that you opened OOo before hand? eg: If you loaded it
> > up then you opened OOo, closed it at some point and then try to re-open
> > it, it'll open up much faster.
>
> No. It was fast from the first time opening. It was nearly instant after that.
> Now it is slowwwwwwwww the first time, about 7-10 seconds after that (which
> is still too slow for a program in ram).
There was a thread on emacs being slow. It was related to
some new input methods.
The way to nail this down is to run "strace" with
timing enabled on both FC1 and FC2 and look for differences.
If oowriter is of interest Something like:
strace -r -f -o /tmp/trace-oowriter oowriter
I noted that it spent a lot of time communication with
cups for printing yet I had not yet opened a file or
clicked on a print icon.
I always wonder at an application that does c.600 gettimeofday()
system calls and all I do is Ctl-Q quit. ;-)
there are also a lot of stat calls for compiled perl
that is not handy followed by an open of perl source text.
stat64("...Carp.pmc", .... = -1 ENOENT (No such file or directory)
open("/usr/lib/perl5/5.8.3/Carp.pm", O_RDONLY|O_LARGEFILE) = 5
I never benchmarked perl compiled .vs. src start up time but there is
a possible opportunity for a speedup in the perl support. I also
noted that kerberos was involved at a lot of places and I believe
kerberos is needed now when it previously could be excluded.
Anyhow compare and contrast.... YMMV.
--
T o m M i t c h e l l
In the USA, vote informed, second Tuesday Nov 2004.
