Re: Cannot typeset tmac/groff_mdoc.n

[Krzysztof Żelechowski]

Dnia wtorek, 25 maja 2010 o 17:37:27 Werner LEMBERG napisał(a):

> > Now to the output of -Thtml: [...]
> 
> You've hit one of the weakest points of groff...

It is also (potentially) the most used point nowadays, as GUI systems prefer 
to display manual pages in HTML: PostScript is too heavy and the X display, 
while most accurate, does not allow to search or copy text.  This is not a 
problem for small pages; however, for pages like groff_mdoc.n it is 
overwhelming (no page references in ToC, no concept index, no keyword index).

> 
> Unfortunately, the author of grohtml no longer takes care of his
> `baby', and I don't have time either to work on it.  For this reason
> grohtml is still tagged as `beta code'.

After giving this problem a second thought, I think grohtml is a 
misconception.  It is an attempt to reconstruct a high-level semantic 
description (HTML) from a low-level graphic description (groff output).  This 
plan is bound to fail.

OTOH, I think it would be possible to modify existing mdoc sources to produce 
semantic HTML directly via write or preferably via device-specific output 
instructions; however, I am not familiar with groff output format so I cannot 
tell whether that would be fully possible, but I know one thing for sure: 
groff has no way to predict the font metrics in the browser that displays the 
result so the output must be high-level to be predictable.  And I envision a 
construct-by-construct transformation to XML as a valuable intermediate step: 
construct-by-construct is easy and XML-to-HTML is easy.

Please forgive me if these ramblings show only my complete lack of 
understanding of the internal mechanisms of groff; it is a large and 
complicated engine and I would obviously need a couple of weeks to understand 
how it works.

> 
> Additionally, the mdoc package has, contrary to ms or man, not been
> prepared for proper grohtml support yet.  On the other hand, there are
> still some bugs directly in grohtml.
> 
> In case you are interested in the latter, it would be of great help if
> you can construct minimal documents which exhibit erroneous behaviour,
> and submit them as individual bug reports to Savannah so that at least
> the problems get collected at a central place.  Submitting patches of
> course is highly welcomed too (for example, to check the HTML stuff in
> `s.tmac' or `an-old.tmac' and do something similar in `doc.tmac' and
> its related files).

I think the bundled BUGS document should mention the Savannah URL.

Cheers,
Chris