mailing-list for TeXmacs Users

[Robert Lamar <address@hidden>]

Hello, TeXmacs users!

This is being written after some investigations following the recent 
"Comparing versions of TeXmacs" thread.  I would particularly appreciate 
input from those who might be in charge of the TeXmacs packages in the 
repositories of Ubuntu and Fedora.  Sorry for a long explanation; I did 
 not want to leave out any clues.

Background: between two computers, I have six TeXmacs installations:

C1: Fedora 7 (Recently upgraded; long story)
(a) TeXmacs-1.0.6.12 (as supplied by Fedora, I believe)
(b) TeXmacs-1.0.6.12-static-gnu-linux
(c) TeXmacs-1.0.6.10-static-gnu-linux
(d) TeXmacs-1.0.6.7-static-gnu-linux

C2: Ubuntu 7.10 (Gutsy Gibbon)
(a) TeXmacs-1.0.6.10 (as supplied from Ubuntu repositories)
(b) TeXmacs-1.0.6.12-static-gnu-linux

The 'static-gnu-linux' variants were all from

  ftp://ftp.texmacs.org/pub/TeXmacs/targz/


My research group has a TeXmacs plugin which was originally developed 
under Fedora 5 with TeXmacs 1.0.6.7.  After upgrading to Fedora 7, it 
ceased to work properly.  The main symptom is this: the plugin passes 
data back to TeXmacs which should look like this:

  ... (tuple 12 "") (tuple 13 "") ...

But under certain installations, TeXmacs reports receiving data looking 
like this:

  ... (tuple #{12}# "") (tuple #{13}# "") ...

And gives the error:

/u1/pg/rl47/.TeXmacs/plugins/mathlang/progs/mathlang.scm:111:21: In 
procedure memoization in expression (for-each (lambda # #) errors):
/u1/pg/rl47/.TeXmacs/plugins/mathlang/progs/mathlang.scm:111:21: Mixed 
definitions and expressions in (begin (define errb (quasiquote ((unquote 
error) (nothing)))) (menu-extend mathlang-error-menu (unquote errb))).

In all cases, the plugin reports *sending* the same (correct) data, but 
in some settings TeXmacs reports *receiving* the wrong (munged) data.

The interesting thing is that this error is seen exactly on the 
installations that were supplied by the Linux distribution (C1-a, C2-a), 
and I get the expected values for the others (C1-(b,c,d) and C2-b).  It 
is not a matter of "before or after 1.0.6.x", and it is not a question 
of the Linux distribution.  It seems like some distro-dependant code is 
interpreting things wrongly. I make no changes to the plugin from test 
to test.

Thank you for reading this far.  Is there anyone here who has an idea of 
what is going on?

Gratefully,
Rob

--
Robert Lamar
PhD Student
http://www.macs.hw.ac.uk/~rob/