Re: Ann: Jörg's LaTeX mode r100

["G. Milde" <milde@xxxxxxxxxxxxxxxxxxxxx>]

On 26.04.07, Jörg Sommer wrote:
> G. Milde schrieb am Thu 26. Apr, 12:06 (+0200):

> > * After evaluating latex.sl, I get a *traceback* buffer with 18 lines of
> > 
> >      ***string***:1: 1 object(s) left on the stack

> Yes, I know. I use stkcheck, but I also use the stack. Is there a way to
> temporary disable stkcheck?

Did you try

  disable_stack_check();
  %stack using code
  enable_stack_check();

But actually, I'd rather use a local or private variable instead of the
stack. Saves a lot of headache and makes the code easier to read (especially
for otheres).

> > * latex.sl
> > 
> >  + If there is an error in the latex run, I get an error
> >      (compose) finished with error

> Yes, you must hit ^Cy to get the log. I find it very annoying, if the log
> window pops up while I'm typing. 

Agreed.

Maybe latex_external->pop_log_file() could get a mode menu entry, so
that I find out the keybinding (which is ^Ey here, due to the CUA
emulation).

> Therefore, only this error message should be in the message line. Or
> what else did you expect?

* Something helpful ;-)

* I'd rather expect a message than an error, because this is no Jed
  problem but a latex error. (My assumption was that some bug in the
  code prevented the correct handling of latex errors.)

* Maybe one could extract and print the latex error message, or the
  offending line number.
  
* A small hint like 

    vmessage("press %sy to see the log", _Reserved_Key_Prefix)

  might be another alternative.
  
> I don't know, how to prevent this backtrace.

Use a message() instead of an error() (or catch the error if it is
needed for aborting a command...)
  

> >   + The rule
> >       dfa_define_highlight_rule("[ \t]+$", "trailing_whitespace", name);
> >     leads to black bg for trailing whitespace if the color
> >     "trailing_whitespace" is not defined.

> Then your color scheme is broken.
> % grep -l trail svn/jed-upstream/lib/colors/**/*(.)

It is local and not updated with the "automatically generated" definitions.
This explains why the custom_color() initialization does not work.

Anyway, as (at least IMHO) the blinking whitespace during typing is
*very* distracting, I'd prefer this syntax rule commented out.

> >   + It would be nice, if latex-external>view() would check if the output
> >     file is still up-to-date and run compose() if not.

> How should I check if the output is up???to???date? 

in rst.sl I use the construct

   if (file_time_compare(buffer_filename(), outfile) > 0)

> Use rubber for this and hit ^Cc before ^Cv. And I think that viewing an
> outdated document is a valid function. 

IMO viewing an outdated document is less often needed than viewing an
updated document. This is why I'd prefer to use a file-listing (or a
shell command) for the outdated output and ^Ev for the up-to-date one.

Would it be save to bind "latex_external->compose;latex_external->view"
to a key or should I expect problems due to the asynchroneous nature of
the compiling.

> Do you use source specials?

No.

Unfortunately, evaluating latex.sl now breaks my jed.

  > xjed -l latex.sl
  Speicherzugriffsfehler
  
thanks again

Günter

--------------------------
To unsubscribe send email to <jed-users-request@xxxxxxxxxxx> with
the word "unsubscribe" in the message body.
Need help? Email <jed-users-owner@xxxxxxxxxxx>.