Re: [Axiom-developer] Re: Pamphlets and LaTex

[Stephen Wilson]

C Y <address@hidden> writes:
> You mean my original code vs. the finite state machine?  The original
> code was almost certainly neither fast or flexible, so it's out.  Of
> more interest is Steve's new work.  I don't know how it does on speed
> trials.  Steve, have you had a chance to run any benchmarks?

Only a simple file with repetitive chunk structure, ~20000 lines long
with approx 800 chunks.  The chunk structure is flat, rather than
deeply nested and hierarchical but that should not matter (until I
implement a walker for the chunk graph to check for circularity's).

So my non-optimized code weaves it in 1.3 seconds and tangles it .6
seconds. noweb takes ~5 seconds for both operations.  I didnt check
cl-web yet, because I cant use an unmodified version with gcl. I did
look at its speed on other examples I am sure it works faster than my
code.  On the smaller examples the speed difference was within a tenth
of a second, neither two appeared slower on average sized files.

For the 20000 line example, LaTeX and GCL each take over one minute to
generate the dvi, or object file.

So really, for normal sized files we are talking only tenths of
seconds, and for huge files we are talking only a second maximum gain
in speed.  In this context, I would rather trade 100 lines of slightly
slower clean code for a faster 400 line tagbody of gotos any day.


Take care,
Steve