-----BEGIN PGP SIGNED MESSAGE-----

Am Sun, 06 Jun 1999 schrieb Andreas Pour:
> Mario Weilguni wrote:
(...=
> 
> You are right, but it does not take 15 seconds.  And also a Pentium 100 is *not* 5
> times slower than a P-II 300 in this case, b/c most of the time is waiting for
> disk accesses to complete, processing time is minimal.
> 
> Here are some better benchmarks, b/c a good part of the delay is due to the fact
> that the bash script is not so fast.
> 
> Just loading the xml file and displaying it (and this dialog is many times more
> complicated than the kppp dialog):  about 0.5 seconds.
> 
> Opening a sub-dialog:  instantaneous.
> 
> The bash bindings are somewhat slow, it is the nature of scripting; bash has to
> load and parse a bunch of definitions, generate some XML and pass it, and this is
> not very fast (python or perl are much faster).  The reason I picked the example I
> did is b/c it pretty much is a worst-case (I guess you did not realize this :-)
> ), but it is still pretty fast.
> 
> Opening a reasonably-sized XML file with only one "page" (like the initial kppp
> dialog):  less than 0.2 seconds.
> 
> > Do you have any idea how fast it is from C++ say for a dialog with the
> > complexity of kppp's setup dialog? If it's slower than 0.2 seconds this is no
> > option.
> 
> There really is not a perceptible difference as far as I can tell.  It all has to
> do with disk access and X drawing times, the actual processing time is not
> noticeable.

Yes, this sounds reasonable. Reading your mail I just got the impression that
the slow part is the XML parser or the dialog-creating part of your library. If
bash is involved, that makes it clear why it's that slow.

Mario

- --
Machine-Independent, adj.:
	Does not run on any existing machine.

-----BEGIN PGP SIGNATURE-----
Version: 2.6.3i
Charset: noconv

iQB1AwUBN1rqImBXd2mp+TWZAQGehAL8D+28oawn1OoXWhvmxDzsJVwna0Rpfj8l
BYqG6MeV28/Jec+cwUxr2IOmos1eUho99apWL7vPYh7hROnujamtagRo9iK6ifmU
eogqrm+ccfH0ngohW0ivIIv4Z04W+hYj
=r+u8
-----END PGP SIGNATURE-----