Enrico Forestieri wrote:
On Mon, Oct 23, 2006 at 11:02:02AM +0200, Georg Baum wrote:
Jean-Marc Lasgouttes wrote:
Good. Does it apply to 1.4 too, or is it an unicode problem?
It is an unicode problem: If char is signed, then the following will produce
an invalid character:
char c1 = 185; //¹ SUPERSCRIPT ONE (in latin1 and ucs4)
lyx::char_type c2 = 185; //¹ SUPERSCRIPT ONE (in latin1 and ucs4)
lyx::char_type c2 = c1; // invalid, not SUPERSCRIPT ONE
Note that there still are problems with char signedness. I am currently
forced to compile with -funsigned-char to avoid crashes when loading
a document on Cygwin. Moreover, I cannot spot them, because they are
hard crashes bailing out of gdb, even.
I reckon this is because Lars reverted my changes vector <-> docstring
in mathed. But Andre was supposed to have a look at that.
Abdel.
