Re: ✘gpsd .23.2~rc1

[Gary E. Miller]

Yo James!

On Sat, 9 Apr 2022 12:40:44 -0700
James Browning <jamesb.fe80@gmail.com> wrote:

> On Sat, Apr 9, 2022 at 12:25 PM Gary E. Miller <gem@rellim.com> wrote:
> 
> > Yo James!
> >
> > On Sat, 9 Apr 2022 12:19:09 -0700
> > James Browning <jamesb.fe80@gmail.com> wrote:
> >  
> > > There are no eternal guarantees. Tomorrow some will probably copy
> > > it to a wrong system and expect help.
> > > Today, it's me. Just as a toy, consider the following.
> > >
> > > $ head -n 50 splot.c
> > > #include <string.h>
> > >
> > > RET faux(int errnum, char *strerrbuf, size_t buflen) {
> > >         return strerror_r(errnum, strerrbuf, buflen);
> > > }
> > > $ gcc -Werror -DRET="int" -c -o splot.{o,c}
> > > $ gcc -Werror -DRET="char *" -c -o splot.{o,c}
> > > splot.c:4:9: error: incompatible integer to pointer conversion
> > > returning 'int' from a function with result type 'char *'
> > > [-Werror,-Wint-conversion] return strerror_r(errnum, strerrbuf,
> > > buflen); ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> > > 1 error generated.  
> >
> >
> > What libc?
> >
> > I'm about to filing a bug report on this with musl.
> >  
> 
> libsystem on expired macOS. I intended for it to be a possible
> standard way to check function prototypes at compile time.
> If the number of successful compiles is one the victor has the type,
> otherwise, someone is messing with you.
> Dig out the logic probes and light stricks. I can occasionally file
> my own bug reports.

I sure hate to have to code an scons test to do that, but may be the
only option.  What worries me is that it is compiler dependent.

RGDS
GARY
---------------------------------------------------------------------------
Gary E. Miller Rellim 109 NW Wilmington Ave., Suite E, Bend, OR 97703
        gem@rellim.com  Tel:+1 541 382 8588

            Veritas liberabit vos. -- Quid est veritas?
    "If you can't measure it, you can't improve it." - Lord Kelvin

pgpjyiZ9qYAeD.pgp
Description: OpenPGP digital signature