[tex-k] Another quick fix

Tomas Rokicki rokicki at gmail.com
Sun Oct 27 00:14:07 CEST 2019

ssize_t is not in the C standard.  I'm not saying we shouldn't use it, but
it might limit portability.  POSIX uses it (and defines it); I'm not sure if
we are limiting TeX to POSIX systems or not.

My feeling is, if we fix any we likely should go through and fix them all
(at least within one subproject like kpathsea),  and then go through the
testing process for all at once.


On Sat, Oct 26, 2019 at 3:02 PM Doug McKenna <doug at mathemaesthetics.com>

> In the file "tex-make.c", in the routine maketex() at or about line 366 is:
>   while ((num = read(childout[0],buf,sizeof(buf)-1)) != 0) {
> Again, the read() function's prototype in the (MacOS/Darwin) header on my
> system says it returns a ssize_t, but because num is an int a compiler
> warning ensues.
> But num is just a local variable, so changing its declaration earlier (at
> or about line 358) to a variable of type ssize_t shouldn't affect anything
> except to get rid of the compiler warning.
> Or does read() deliver different sized integers on different systems?
> Doug McKenna

--  http://cube20.org/  --  http://golly.sf.net/  --
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://tug.org/pipermail/tex-k/attachments/20191026/1f91f019/attachment-0001.html>

More information about the tex-k mailing list