Re: [Texmacs-dev] Re: TeXmacs+Axiom

[Joris van der Hoeven]

On Tue, May 22, 2007 at 12:00:34PM +0200, Ralf Hemmecke wrote:
> >when putting large pieces of the document in certain markup like tables,
> >or having a document of several hundreds of pages.
> 
> Oh, I probably would like a document, that is like a master file with 
> hundreds of include files. I all want to see them as *one* (hyperlinked) 
> document. If you now say that TeXmacs will be slow in this case, I know 
> where I have to stop.
> 
> Remember for me it is not important that the 1000+ pages are always what 
> I see when I print it. In fact, I probably will never print them. It is 
> more important to have the hyperlinks (approximately) correct and the 
> screen output reasonably nice. So a button to re-beautify would be 
> totally fine for me.
> 
> I know, I enter an area here (literate programming) for which TeXmacs 
> was probably not designed, but I wanted to see whether it would be 
> reasonable to consider TeXmacs. Speed *is* an issue for almost any 
> programmer.

Sure, but TeXmacs has been designed to be fast for "normal" documents
of a "normal" size. Typically, on a modern computer, TeXmacs typesets
about 10 to 50 pages of mathemathics per second, which makes loading
sufficiently fast. When making local modifications, the typesetting
is redone only for the current paragraph.

So hyperlinked documents which don't exceed 10 pages individually
will typically be very fast.

> >>Oh that is a big issue. I haven't yet used TeXmacs again extensively, 
> >>but this slowness in keyboard response was one of the reasons I turned 
> >>away about one or two years ago.
> 
> >Examples, timings?
> 
> How do you expect me to messure the timing between pressing a key and 
> seeing on screen until nothing moves anymore? Of course, I understand 
> that you want to see timings, but I don't know how to give them. 
> Otherwise I would probably have sent a message two years ago.
> 
> But now I should stop and actually test TeXmacs, maybe things have 
> change (like me having a faster laptop ;-) )

In order to be able to do something about a remark "I feel that
TeXmacs is slow", I need more information.

There are essentially three important efficiency issues:
  1) Booting speed.
  2) Brute typesetting speed.
  3) Display speed.
So one first has to make precise about which speed one is talking and
then give me detailed and reproducible information about what action
takes how much time. Of course, as Henri pointed out, the third point
may depend on your graphics card, so reproducibility may be hard
in that case...

> >>Hi, TeXmacs developers, is there a trick to delay expensive computations 
> >>until I press a button "re-beautify"?
> 
> >Rendering is already done in that way.
> 
> Could you detail a bit? Or where do I find a description of how the 
> delay process works. Please don't let me read code, I will probably not 
> understand the essence.

Basically, local retypesetting is done after every change.
When typesetting is done, we render, text close to the cursor first.
The rendering process is interrupted on a new keyboard event,
even though a minimal number of boxes are always redisplayed.

I tested this strategy on several computers and installations and
it seems to be the best possible. The only thing which can still be
essentially improved is locality of the typesetting process.
Nevertheless, for an ordinary paragraph, a complete typesetting
is of the order of 0.01ms, so this is not the main bottleneck.

As to strange graphics cards, one would have to investigate...

> BTW, is there an SVN archive of TeXmacs?

We still use CVS: see Download on the website.