Re: struct timespec on NSK (FYI)

[Matthew Woehlke]

Paul Eggert wrote:
> Matthew Woehlke writes:
> > While trying to build gzip 1.3.7, I noticed inconsistent usage of
> > gnulib's timespec.h via stat-time.h, which led me to dig up where
> > 'struct timespec' is defined on NSK. It turns out the declaration is
> > in spt_types.h, included in spthread.h.
>
> Does this actually matter?

I'm not sure it does. I was going to say so in my original post until I 
looked up the actual declaration.

> That is, does the gnulib (or gzip code)
> use any system functions that use the spt_types.h version of struct
> timespec? 

...and this is the real question. Realize, by "the spt_types.h version" 
you mean "struct timespec" (actually, it turns out that pthreads.h would 
pick up one in dec/cma_px.h - who woulda thought? - that is the same), 
since this is the ONLY variety of "struct timespec" I could find on NSK. 
One would have to assume "no", since no header defining it was pulled 
in, although the call to localtime() which uses the tv_sec member is 
more likely to be problematic. Except, ironically, time_t is correct for 
that. :-)

Other than that I don't see gzip using it.

> If not, then let's not worry about it: gzip will use its
> own struct timespec which conforms to POSIX, and this will work better
> for gzip than trying to deal with a nonstandard version of struct
> timespec that doesn't really matter.

Given the difference (tv_secs signed vs unsigned), do you really expect 
treating a signed as an unsigned to matter? Is tv_secs allowed to be 
negative?

> > There might be a problem with the replacement, since NSK's types.h
> > defines time_t as 'long', but spt_types.h gives the type of tv_sec as
> > 'unsigned long'.
>
> Sounds like a recipe for bugs to me.  Let's avoid the NSK struct
> timespec if we can.

gzip looks OK (sans the problem in zip.c). Not sure about the rest of 
gnulib; does it ever pull any of the following headers?
spt_types.h cma.h cma_*.h pthread.h spthread.h pthread_exc.h
(pthread.h is the only one I might expect to see, the rest I think are 
probably NSK-specific.)

If not, then it is not going to pick up NSK's struct timespec, therefore 
it will not pick up any system functions using it.

--
Matthew
HIPPOS wallow slightly in the MUDDY RIVER
What do you want to do next?
> WALLOW IN MUDDY RIVER
You join HIPPOS.