On Wed, Sep 18, 2002 at 12:44:06PM +0200, Andre Poenitz wrote:
 
> On Mon, Sep 16, 2002 at 09:36:24AM +0300, Martin Vermeer wrote:
> > While we're at it (and this is a apropos the textsf bug too) why
> > (mathed/formulabase.C):
> > 
> >     //  Math fonts
> >     case LFUN_GREEK_TOGGLE: handleFont(bv, cmd.argument, "lyxgreek"); break;
> >     case LFUN_BOLD:         handleFont(bv, cmd.argument, "textbf"); break;
> >     case LFUN_SANS:         handleFont(bv, cmd.argument, "textsf"); break;
> > [...]
> 
> This is the handling of LFUNs.
> 
> > but (lib/symbols):
> > 
> > mathnormal        font        mathmode  inherit  medium   up  math
> > frak              font        mathmode
> > mathbb            font        mathmode
> > mathbf            font        mathmode
> 
> This is the handling of "direct keyboard input" like
> 
>  \mathbb
> 
> Ideally, the should be the same. However, there are not.
> 
> > I found experimentally that the font names used in formulabase.C must be 
> > present in lib/symbols in order for LyX not to go haywire.
> 
> Indeed.
> 
> > Questions: Why in math use text font commands? 
> 
> Because math can contain nested text in \mbox, \fbox, \textrm etc.
> 
> > And why are some of those in lib/symbols and some not?
> 
> Because nobody tried to come up with a scheeme where this kind of
> inconsistencies are not possible.
>  
> > Anyway, replacing textsf by mathsf and textnormal by mathnormal in
> > formulabase.C fixes both your problems... but is this a proper fix?
> 
> Well. The question is "What is LFUN_SANSSERIF" supposed to do. My
> implementation answered this with "switch to \textsl", yours says "\textsl
> if nowhere close to math, but \mathsl if somewhere within math, even
> within nested text boxes".
> 
> I actually believe we should replace all the different LFUN_FONTCHANCES by
> a single LFUN_FONT as a font is nothing that should be hardcoded...
> 
> Andre'

I see. Explanation accepted. So what you're saying is that we actually
should have buttons for both \mathsf and \textsf within math (no matter
how implemented)? 

Anyway, I suggest you do something that fixes the 'haywire' behaviour
with a minimum of collateral damage. Add textsf to symbols for all I 
care :-)

Martin

Attachment: msg44877/pgp00000.pgp
Description: PGP signature

Reply via email to