Re: [MiNT] MiNTlibs for PureC

[Guido Flohr <gufl0000@stud.uni-sb.de>]

Hi,

On Wed, Sep 08, 1999 at 03:25:39PM +0200, Helmut_Karlowski@mausd.gun.de wrote:
> Hallo Jo
> 
> JES>Yes. And unless Guido has done something with the makefile since I
> JES>downloaded 0.51 you'll have to hack it too. It took me about 15 minutes
> JES>to hack the makefile, the build-process took less than five minutes
> with
> JES>PureC on my Afterburner... With Gcc it takes easily 30 minutes.

Hm, I *have* hacked the Makefiles.  But if you have both gcc and Pure it
should be easy to update these Turbo C project files.

> First Bug (reported here for convenience):
> 
> In tz.c, #165:
> 
> ---------------><---------------
>     /*if ((fid = open (name, O_RDONLY)) == -1)*/
>     if ((fid = open (name, O_RDONLY)) < 0L )
> ---------------><---------------
> 
> If 'name' doesn't exist, open returns -4, not -1.

That would be a bug in open.c not in tz.c.  I'll check that.

> In new_york.h the constants need an 'L' (caused warnings):
> 
> ---------------><---------------
>   3,  /* typecnt */
>   12,  /* charcnt */
>   { /* ats */
>     -1633280400L, -1615140000L, -1601830800L, -1583690400L,
> ...
> ---------------><---------------

OK.

Just btw,  these constants in new_york.h belong to an initializer for a
structure, a smart compiler should actually know from the declaration
which types the constants have.  You ANSI C experts out there: What does
the standard say?

I mean, if they were not "long" but char, the compiler would also have to
know that a "1" is not an integer or long but a char.

Ciao

Guido
-- 
http://stud.uni-sb.de/~gufl0000/
mailto:gufl0000@stud.uni-sb.de