Subject: Re: porting problem
From: Mike Nordell (tamlin@algonet.se)
Date: Sat Jan 27 2001 - 21:31:15 CST
Hollis R Blanchard wrote:
> On Thu, 18 Jan 2001, Sam TH wrote:
> >
> > ICONV_CONST is a horrible hack that I created to fix this problem of
> > incompatible iconv() prototypes. What you need to do is find where
> > ICONV_CONST is defined, and add something like this:
> >
> > #ifdef __MAC_OS__ // or whatever your platform defines
> > #define ICONV_CONST const
> > #endif
> >
> > That should fix the problem.
> >
> > This is a hacked-up way to get around our lack of configure, which
> > shold evetually be remedied (once my computer works again).
No! This is *NOT* "a hacked-up way to get around our lack of configure".
This is a BUG, and in the most serious sense, in the iconv interface, and we
apparently have to suffer from it.
I'd say we just wrap it in our own forwarding function that "just fixes it".
For the platforms where it takes non-const data we perhaps should take a
copy of the input data and assert equality on return (though I think that
would be to be overly "careful", but since iconv is apparently designed and
implemented by "uncareful" developers perhaps it's better to be safe than
sorry?).
> P.S. Is it "abiWORD" or "abiSUITE"? A lot of confusion there.. RH ships
> abiword, you guys build abisuite...
Not at all. We build AbiWord. AbiSuite was (is?) the vision of creating more
than one "office-type" application on the same framework; the ABI framwork
(which is *not* Application Binary Interface, however much you'd like it
:-) ). At least that's how I got it.
/Mike - don't cc
This archive was generated by hypermail 2b25 : Sat Jan 27 2001 - 21:30:45 CST