[Axiom-developer] RE: windows/linux coherency

[Bill Page]

On Thursday, December 23, 2004 10:40 AM address@hidden
wrote:
> Bill Page wrote:
> > 
> > If the new tm_axiom consults the TM_AXIOM environment variable
> > and finds
> > 
> >   TM_AXIOM=texbreaker:n
> > 
> > then it can be made to behave exactly the way the cygwin
> > tm_axiom does now.
> 
> Great! I downloaded the latest version and it reads OK the 
> parameter value.

Note of course that the syntax has changed a little to
conform to the Axiom )set output command. So the above
should be written:

TM_AXIOM='breaker no'

> Maybe it would be useful to add in the doc that it is necessary
> to got out of TeXmacs and rerun it, to take into account a
> modification?

Ok, thanks.

> 
> Unfortunately, it seems that it is still necessary that 
> axiomsys.exe be either in the PATH or in the
>   texmacs/plugins/axiom/bin
> directory. I succeeded in making nevertheless
> Axiom run inside TeXmacs and keeping axiomsys.exe in the Axiom 
> tree, by adding an axiomsys.bat in the TeXmacs\plugins\axiom\bin
> directory, then changing the bat extension into exe extension,
> but it's a rather dirty and temporary hack. I think it would be
> better if tm_axiom could find the true binary axiomsys.exe 
> from $AXIOM\bin.
>

tm_axiom already finds axiomsys.exe from $AXIOM\bin. tm_axiom
does not use the PATH.

The problem that you are having is that TeXmacs itself is looking
for AXIOMsys as a sign that it should include Axiom in the
Insert/Session menu. It serves no other purpose so your hack is
not so bad. TeXmacs also looks for tm_axiom in the PATH but
TeXmacs adds
 texmacs/plugins/axiom/bin
to the PATH variable locally before doing the search so if
you move tm_axiom here (as is usual in TeXmacs installations)
then you will not have a problem.

In the windows Axiom installer puts tm_axiom in the
  $AXIOM/mnt/windows/bin
directory since it can not be sure whether TeXmacs is present or
not. The current native Windows version of TeXmacs does not
include any tm_axiom binary, but as long as the PATH includes
  $AXIOM/mnt/windows/bin
then everything works.
 
> I am sorry to insist on having (if possible) axiomsys.exe NOT 
> run from the PATH.  I know it's rather orthogonal to the usual
> practice on Unixes, but on these systems there is a superuser
> that is supposed to "know" and install everything cleanly. On 
> Windows, users often install themeselves, and they can install
> different versions or different distributions of a given program,
> so that, if several binaries with the same name are in the PATH,
> it can produce very strange results. This is a classical 
> problem e.g. for TeX distributions (and I solved the pb in 
> AsTeX in this way, so that it is compatible with all other
> TeX installations, even if they use the PATH).

I understand. That is not a problem I said above. To correct
this behaviour requires a change to the axiom_init script in
TeXmacs so that it not longer looks for AXIOMsys. Maybe it
should look for the AXIOM environment variable instead. But
right now I do not know enough about scheme scripts in TeXmacs
to make even this simple change.

If
  $AXIOM/mnt/windows/bin
is not in the PATH, then right now I can only think of one
minor problem that would affect the way AXIOM works. The current
version of AXIOM for Windows expects to be able to run a program
called 'cat' (after the unix command of the same name), it order
to display some help files. In the current version there is a
simple 'cat.exe' program in 
  $AXIOM/mnt/windows/bin
that does the job. I think you will have to move this to
  texmacs/plugins/axiom/bin
so that the )summary command will work in your CD configuration.

> 
> - For the "l" and "c" parameters in Sutor's output, these are 
> the "left" and "center" parameters of the array environment.
> In Sutor's output, the "l" is used, but it is rendered centered
> inside TeXmacs, as if the "c" parameter was used instead, so I 
> suspect it is a bug in TeXmacs import filter.

Are you familiar with how to file a bug report for TeXmacs.
If not, then I can do this later.

> 
> - For the output with nested arrays (in the example with -1/3 
> power), the structure is the following:
> 
> begin array
> - 
>    begin array 
>      text
>     \\
>     \\
>     text
>   end array
> text
> end array
> 
> I think it ought to be either
> 
> begin array
> - 
>    \left( begin array 
>      text
>     \\
>     \\
>     text
>   end array\righ)
> text
> end array
> 
> to be correct, and I supposed that, if the original form was 
> intended (i.e. with no \left( and \right)) then probably it
> ought to be
> 
> begin array
> - 
>    begin array 
>      text
>     \\
>     text
>   end array
>     \\
> ext
> end array
> 
> i.e. the second \\ in the first array would seem to be 
> incorrectly placed.
> 
> Another reason I would suppose the original form is not what 
> was intended : the motivation for using a LaTeX array seems
> to be to break a long line into several small ones. So :
> 
> - if one breaks a long part into 3,
> - and then adds the sequel after the block of 3 lines on the
> same line, it seems rather contradictory ?  But of course,
> it is just a supposition.
> ... 
> And a last point about axiom.ts : I think there might be a 
> problem on Windows : the type is still written just after the
> number of the formula, not on the next line. Is your modified
> axiom.ts supposed to write it on the next line, or did I
> understand you incorrectly? (sorry, I don't write from my PC,
> I answer just from memory - it might be deficient :-))
> 

Yes, that is what the modified axiom.ts does. It forces the
Type: information to a new line. In order to use it, you
*must* replace the existing axiom.ts in TeXmacs with the
new one manually copied from
  $AXIOM\mnt\windows\src\texmacs
To just install AXIOM for windows alone is not enough. I do not
know how to make this automatic.

Regards,
Bill Page.