On 29/10/2008 20:15, Enrico Forestieri wrote:
On Wed, Oct 29, 2008 at 04:40:13PM +0100, Enrico Forestieri wrote:
On Wed, Oct 29, 2008 at 04:30:41PM +0100, Abdelrazak Younes wrote:
So my initial assumption that it is wrong to allow InsetMathHull inside
InsetMathHull was right?
Yep.
I was too quick here (no time to think), see below.
FWIW, the same bug plagues LyX-1.5.
I'll try to have a look later tonight, if nobody beats me to it.
The attached patch makes it impossible to create certain bad math insets.
For example, when selecting $\[x\]$ and then hitting Ctrl+M, LyX will
complain that that is a bad formula.
Unfortunately, there are other cases that are difficult to catch (i.e.,
I don't know how to do it, yet). E.g., this is the case for $\textrm{\[x\]}$
which is still happily accepted by LyX even if it is uncompilable by LaTeX.
OK. But at least the situation is improving WRT 1.5.
Contrarily to what I stated before, it can happen that InsetMathHull nests.
For example, this happens for $\text{$x$}$ which is a valid LaTeX construct.
So, after all, your r27186 was still needed, Abdel. Sorry for having
requested its reversion. Please, reapply it.
No problem. Stefan's patch looks more promising anyway so I'll wait a bit.
Abdel.
