Re: [Groff] new grohtml was (HTML fonts)

[Werner LEMBERG]

> I like the idea of the markers - this will help .br .in .ce etc but
> the biggest problem so far in current grohtml is the textual table.
> Will the above solve this?

I think so.  With `raw data' I mean the input troff sees *before*
processing with the tbl program.

So

  ... | tbl | groff -Thtml ...

will fail but

  groff -t -Thtml ...

will succeed.  The only alternative I see is to divide grohtml into
two parts, a preprocessor (say, pre-html) which must be called before
tbl, and grohtml, the output device:

  ... | prehtml | tbl | groff -Thtml ...

Not very user-friendly IMHO.

Another idea: What about modifying tbl so that it directly produces
HTML tables (or inserts appropriate markers)?  After some thinking I
believe this is the most promising way to get good results, and the
abovementioned problems vanish.

> What is grohtml stage one doing, my understanding is that it:
> 
>  (i)  reads in raw troff input - adds say \X html-ce
>       \X html-br etc etc for all the markers.
>  (ii) produces raw HTML using by correlating the raw troff
>       input with the troff output ??

Not really; with `raw HTML' I mean everything which is fully
representable as HTML using -Thtml.

>  (iii) raw + markers get sent to groff -Tps
>        to generate images
> 
> then grohtml stage two
>  (i)  checks to see whether the troff output (needs troff
>       output -Thtml as well?) can produce simple eqn's etc
>       without using -Tps ?
> 
>  [ Actually can't grohtml stage 1 decide
>    whether the eqn etc can be produced using html gylphs
>    rather than stage2 - unless I've misunderstood something :-) ]

Of course this can be done in stage 1 also.


    Werner


PS: Do you know latex2html?  Since LaTeX tables are defined quite
similar to `tbl', you may get some ideas from it...