Re: [Texmacs-dev] Explanation for patch #2590

[Nix]

On Thu, 2004-02-12 at 06:38, David Allouche wrote:
> > ((code (document "The Truth")) "The Truth is not so simple.")

Sorry, forgot the double quotes - corrected.

> 
> That is not a proper texmacs document snippet in Scheme format... well
> maybe that's something used internally in the tmtext converter, but I am
> not familiar with its internals, so I cannot really tell what that
> represents and whether that's a valid texmacs document snippet to start
> with. In any case, the nested "document" looks weird to me.

Save the previously attached TeXmacs document as Scheme.

> I am not sure what the gluing stuff is intended to do (not that I have
> checked recentely). I have even made a point previously that designing
> multiple pass converters in that way lead to hard to read code because
> one has to understand the whole conversion process in order to
> understand any given pass.
> 
> Clearly, the intent is not to introduce sectioning markup. Maybe the
> problem is it inserts "paragraph" tags instead of "para", and thing
> breaks since "paragraph" are now properly handled as sectioning markup.

The alternative to removing the gluing code is to prune the 2 DRDs used
to decide whether things are glued together using the code.  However, I
have come to the conclusion that pruning the DRDs in this fashion would
leave them empty.  IOW, we might as well remove that code.

Inserting "para" tags would be an extremely bad idea, since they refer
to header/footer stuff, which regular paragraphs are not.

To be perfectly honest, I don't see the point of that code at all.
Regular paragraphs are represented in TeXmacs as simply a list.  IOW, we
have

(document "This is the first paragraph." "This is the second paragraph"
...)

Why should there be any code that combines 2 paragraphs into one in
/any/ way ?

I have answered this question by removing the gluing code from
tmpre.scm.  I don't believe it could have served any purpose other than
to insert paragraph tags, because, well, that's all it does.  It
basically goes through the document one paragraph (one list item) at a
time, performs tmpre on that paragraph, and conditionally glues it to
the following paragraph (which has already gone though tmpre).  I have
found that there are /no/ conditions under which 2 paragraphs should be
glued.