Dynamically linking with a Unix library - or not

[Behdad Esfahbod]

On Tue, 18 Jun 2002, Francois Gouget wrote:

> On Tue, 18 Jun 2002, Shachar Shemesh wrote:
> A way to deal with B is to do like for FreeType, i.e. load the library
> dynamically at runtime, and disable support for it if that fails.
> 
> But I have another question which you have probably already envisioned
> (might even have already discussed on the list though I can't remember
> right now): FriBiDi being a Unix Unicode library, I assume it uses
> 4-byte Unicode characters? Won't it be a problem since Wine uses 2-byte
> Unicode characters?
>
> That may be another argument for integrating FriBiDi with Wine, or
> maybe:

Yes, this is what I suggested too.

>  * adding a compile option so that FriBiDi works with either 2 or 4 byte
>    characters

I'm working on it, currently there is no compile time option, but 
you can do it by changing one character from 4 to 2 in a header 
file, and it will regenerate the tables.

>  * make it possible to write a configure check to detect whether the
>    FriBiDi library is compiled for 2 or 4 byte characters

It's already possible, something like this:

AC_TRY_RUN ([
#include <fribidi/fribidi_types.h>
int main() { return sizeof (FriBidiChar == 2) ? 0 : 1; }
], FRIBIDI16=true, FRIBIDI16=false
)

>  * then packagers would compile Wine with this special FriBiDi library,
>    and ship it with Wine
> 
> How does that sound?

The right choice.

> --
> Francois Gouget         fgouget at free.fr        http://fgouget.free.fr/
> May your Tongue stick to the Roof of your Mouth with the Force of a Thousand Caramels.

-- 
Behdad Esfahbod				29 Khordad 1381, 2002 Jun 19 
http://behdad.org/			[Finger for Geek Code]

Proof techniques #1: Proof by Induction.

This technique is used on equations with "n" in them.  Induction
techniques are very popular, even the military used them.

SAMPLE: Proof of induction without proof of induction.

        We know it's true for n equal to 1.  Now assume that it's true
for every natural number less than n.  N is arbitrary, so we can take n
as large as we want.  If n is sufficiently large, the case of n+1 is
trivially equivalent, so the only important n are n less than n.  We
can take n = n (from above), so it's true for n+1 because it's just
about n.
        QED.    (QED translates from the Latin as "So what?")