Re: [Texmacs-dev] {under,over}brace work

[Joris van der Hoeven]

> On Wed, 2003-04-16 at 03:44, Joris van der Hoeven wrote:
> > This certainly does have an effect. In math-mode you will
> > be able to get them using C-q.
>
> I'm afraid not. All I got was an ugly, red <braceld> when I typed C-q,
> braceld, <enter> in math mode. My rubber-cmex.enc file has the following
> appended to it:
>
> 122 "braceld" "bracerd" "bracelu" "braceru"

I guess that it should not actually be in the rubber-cmex.enc,
but rather in the cmex.scm.enc (if that one exists).

> Also, this does not solve the problem of the txexa font wherefrom I need
> the brace extension character.  An overbrace consists of the followings:
>                                 braceru   bracelu          bracerd
>                                        v v                        v
>   braceld -> .- ----<-txexa #32->---- -' '- ----<-txexa #32->---- -.
>
> So, basically, I also need txexa.  Here's what I've tried so far:
>
> File txexa.enc:
>
> 32 "bracext"
>
> file math.enc:
> ...
> [txexa]
> ...
>
> All this got me when I typed bracext was a really weird symbol, and it
> didn't say
>
> "Please wait, generating txexa<some-number>.<some-extension>"
>
> Also, since the symbol looked like a <nsucccurlyeq> rotated Pi/2, I
> tried to click on the menu with the <prec> symbol to try to identify it,
> but at that point, TeXmacs crashed with
>
> Fatal error: Font not found in 'math_font_rep::search_font'
>
> Creating a file called cmex.enc with the line
>
> 122 ...
>
> did work, but produced the wrong symbols.
>
> Conclusion:  This is not working.

Are you sure that we cannot find a replacement for txexa #32;
I don't remember having seen this font before...

> > However, making wide under/overbraces work will require
> > one or two days of work. The best thing to do is to take
> > a look at how large delimiters are dealt with in the C++ code.
> > Be careful though that there is no automatic TeX-font information
> > about wide composite symbols (as there is for the large delimiters).
> > You should also pay attention to the tricks to get
> > the anti-aliasing right.
>
> Actually, I was thinking about using virtual fonts and Scheme functions
> such as (glue ...).  I was thinking about making it just another
> character, to be used with (make-wide ...) and (make-wide-under ...).

No; I think that you really have to write something similar as
for the large delimiters. This will both allow you to generate
a potentially infinite number of brace characters and to write
a routine which computes the right brace in a given situation.

> However, if I can't get to the point where I can see these symbols, I
> cannot proceed further on any path :o(

Notice that you can't see the large delimiter symbols
(<large-(-1>, <large-(-2>, ...) either. This is handled
via the C++ interface.

Notice also that the "pointer" of the brace is not necessarily
in the middle; we probably need characters <overbrace-n-m>
and <underbrace-m-n>, where m and n are potentially different.