lout-users
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: self-loops in graphs


From: Michael Piotrowski
Subject: Re: self-loops in graphs
Date: Thu, 18 Jun 1998 21:30:06 +0200 (METDST)

Hi,

First of all, thanks to Jeff for the hint.  I just found the posting (dated 
last Thursday):  I had somehow missed it (perhaps because the subject was 
different from mine) and I complained to a co-worker that nobody seemed to 
care,  when he pointed out that there was indeed an answer to my question on 
the list :-)

It works great.  Here is what I currently have;  maybe someone's interested:

File "mydefs":

   extend @DiagramPrint @Diag
   macro @FinalNode {
      @Circle vmargin{0f} @Circle hmargin{0.5f}
   }

File "diag.lt":

   @Diag
   {
      Start:: @Node outlinestyle{noline}
      @DP
      ||1c One:: @Circle 1 ||2c Two:: @Circle 2 ||2c Three:: @FinalNode 3 //1c
      //
      @Arrow       from address@hidden to address@hidden ylabel{r-1}
      @Arrow       from {One}      to {Two}    ylabel{r-2} ylabelprox{below}
      @CurveArrow  from address@hidden   to address@hidden ylabel{r-1} 
ylabelprox{below}
      @CCurveArrow from {One}      to {Three}  ylabel{r-3}
      @Arrow       from {Two}      to {Three}  ylabel{r-3} ylabelprox{below}
   }

The diagram looks really good, but I'm not yet entirely satisfied with two 
things:

- The way I coded the start state is clumsy, and there might be more than one, 
and then it'll become a mess.  I tried to code the in-pointing arrow as a 
label which contains a diagram which only contains an arrow, but I need to be 
able to label this arrow, and I'm not yet good enough at Lout (I have to admit 
that I find the syntax not really more intuitive than TeX's) to handle this;  
it didn't seem very elegant to me, either.

- The code for the end state seems equally bad to me;  I'm not really sure why 
it works, and I hate that :-)

Since I'd eventually like to have these diagrams directly produced from 
corresponding grammars, I'd prefer a clean solution.  I'd thus be grateful for 
any comments.

Thanks again

-- 
Michael Piotrowski                         <address@hidden>
                              <http://www.linguistik.uni-erlangen.de/~mxp/>
Department of Computational Linguistics --- University of Erlangen, Germany


reply via email to

[Prev in Thread] Current Thread [Next in Thread]