[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Texmacs-dev] Re: [TeXmacs] helping to speed it up

From: Henri Lesourd
Subject: [Texmacs-dev] Re: [TeXmacs] helping to speed it up
Date: Thu, 12 Oct 2006 13:18:43 +0200
User-agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.4) Gecko/20030821

Kasper Peeters wrote:


I am trying to figure whether there is anything obvious that I could
do in order to make interactive CAS sessions be snappier. The input &
output are great, but even though I just got a new core duo machine
and even though the underlying CAS spits out answers without noticable
delay, it takes around 1 second to evaluate each cell in my

What do you call a "notebook" ? A CAS session inside a TeXmacs document ?

is there perhaps a way in which
someone like me, with a decent programming background, can help to
improve the speed?

The first thing to do is to build test cases, to be sure that the slowdown
really stems from TeXmacs. Namely, an example in the context of
a particular CAS, where one can clearly see that the same calculation
outside of TeXmacs (e.g., measured with 'time' under a UNIX shell)
spends much less time that when it is done from inside TeXmacs (for
this, although it is not very difficult to redo, I can provide some Scheme
code for measuring time, if needed).

This way, we would have a benchmarking framework to clearly be
able to agree on what the slowdown phenomenons exactly are.

Are there perhaps parts of TeXmacs which are
already known to be particularly affecting CAS sessions, so that I
could more easily focus on those?

TeXmacs speaks to the CAS by means of some kind of RPC
mechanism, thus it **should not** slowdown that much. On
the other hand, synchronising the changes in the document
can take some time, because you must wait to perform
your changes at the appropriate time regarding the display
process (namely : **after** the previous modify / redisplay
cycle is completely done).

But even this should not slow down things that much, thus
I'm surprised...

reply via email to

[Prev in Thread] Current Thread [Next in Thread]