[help-3dldf] Re: CVS 3DLDF:scanprse.web:451: error: `yyparse' was not declared in this scope

[lfinsto1]

Hongyi Zhao wrote:

>
> Thanks, see the following results, please:
>

That shows that it works.  What happens when you change `LANG' back to
your usual value and run 3DLDF after having installed it with English as
the language?  If it works, I wouldn't worry about this problem any
longer.  Otherwise, you could just change it in the shell in which you're
using 3DLDF.  You could put `export LANG=en_US' (or whatever) in a shell
script, or modify the `Makefile' generated by `configure', or modify
`Makefile.am' and run `autoconf', `automake' and `configure'.  `automake'
and `autoconf' are run in the `3DLDF-1.2.0.0' directory, i.e., above the
`CWEB' subdirectory, like `configure'.  You might have to run them like
this:

autoconf && autoheader && automake --add-missing --copy
configure --prefix=[your path, as discussed before]

I don't know why there's a problem with the template functions when you
set `LANG' to Chinese.  I can't tell from the error messages, because I
don't understand Chinese, and they disappear when you change the value of
`LANG'.   If you want me to try to find out what the problem is and fix
it, you'll have to tell me what they say.  They're probably all the same,
or at least very similar.  I would probably then have to contact the GCC
developers.  Lacking further information, my opinion is that this
shouldn't be happening.  A program should not fail to compile and/or link
due to failures in the Standard Template Library, just because the value
of `LANG' is changed.

If you translate the error messages, or at least the first few, I will try
to find time to work on this problem, if it's important to you.  I'm sorry
to sound reluctant, but I'm having some problems not related to
programming  at present.  However, my advice would be to work around the
problem, as outlined above, if possible.

When I said it would be "risky" to let 3DLDF read arbitrary characters, I
didn't mean this in the sense of it being a security risk.  I meant that
it seems like something that might cause problems for me later with
respect to programming.

Having given it some more thought, I think it probably wouldn't do any
harm.  If I get a chance, I may add another category code: `UNKNOWN_CHAR'
and make the necessary modifications that would allow other characters to
be used in `strings'.  I believe this could be done easily, but would have
to check to be sure.  If they occurred anywhere else, for example, in the
name of a variable, they would cause a parse error.  3DLDF doesn't fail
gracefully when parse errors are encountered.  This is something I'd like
to change, but it's not urgent.  It does no harm to just end the program
with `Control-C'.   In your own sources, it would probably suffice to put
the following lines at the end of `get_catcode' in `scan.web':

else
   return SYMBOL_7;

This would also make characters not otherwise accounted for be valid in
variable names (actually "tags" and "suffixes", to be precise).  However,
I advise against doing this, since this wouldn't be compatible with the
way I will do it, when and if I get a chance.

In addition, you can use LaTeX in labels in 3DLDF in the same way you
would use them in MetaPost.  If you need to write any `verbatimtex' code
to the MetaPost output, you can use `verbatim_metapost' to do so.  There
are examples of this in `sample0.ldf'.  If you're using LaTeX, you
probably know how to include the PostScript files generated by MetaPost in
LaTeX files.  I don't have any examples of this in the 3DLDF package, and
won't be adding any.  I'd have to look up how to do it myself, and it's
not really the purpose of the package.  I just include the TeX files as a
convenience.

Laurence