[tex-live] Re: [tex-pretest] Web2C 7.5.2 -- "critical fixes"

[Olaf Weber]

Fabrice Popineau writes:

>> I'm not sure I understand your point?

> I mean to be exactly symetric, REGFIX should always be defined.

Speed, I think.  It's old code.

>>> EncTeX is dumping a boolean and it was dumped as an int and undumped
>>> as a boolean (different sizes). So be warned. I had to define REGFIX.

>> Don't ever dump booleans directly.  Where is encTeX doing this?

> in enc-option.patch-to-7.5 :

> +dump_int(enc_option);

> If you dump a boolean using dump_int, it will be cast to an int. If you
> undump it without REGFIX defined, then it will be undumped as a boolean,
> at least on my machine/compiler. 

You shouldn't dump booleans at all because on different machines you
can end up with different sizes: note how mltex_p is handled.

There's some other stuff that seems off in the patch: web2c's
'dump_int(hyph_prime)' gets turned into 'dump_int(hyph_size)'; I'm not
convinced that is a good idea.

I may also end up suggesting somewhat different messages.

-- 
Olaf Weber

               (This space left blank for technical reasons.)