On 08 May 2001 20:03:47 +0500, Dan Winship wrote:
> > Uhm, I am confused here. Why cannot we change that code to report an
> > error?
>
> Didn't say we couldn't, just that it would require taking some amount of
> performance hit.
I would suggest this performance hit is negligable (particularly given
all the other processing requirements for rendering utf8, this bizarre
easy to crash interface is totally nuts).
Anyway, many of the cases we're hitting are something like:
for (p = text; *p; p = next_utf8_char(p)) {
}
(or whatever it is exactly)
We can just add a check that p actually advanced without too much effort
(and a pretty small performance hit at that, 1 move and 1
subtraction/branch per loop), and that will fix the majority of the
problems WE are facing (which just pissed me off this mornign with a new
mail that came in in widnows-1254 charset).
Z
_______________________________________________
evolution-hackers maillist - [EMAIL PROTECTED]
http://lists.helixcode.com/mailman/listinfo/evolution-hackers
