uname() in mdoc_validate.c on Solaris

uname() in mdoc_validate.c on Solaris

[Yuri Pankov]

Hi,

uname(2) on Solaris (OpenSolaris, OpenIndiana, Solaris 11 Express)
states:

RETURN VALUES
     Upon  successful  completion,  a   non-negative   value   is
     returned.  Otherwise,  -1  is  returned  and errno is set to
     indicate the error.

Could we possibly change

    if (uname(&utsname)) {

to

    if (uname(&utsname) == -1) {

to make .Os without arguments work on these OSes?


TIA,
Yuri
--
 To unsubscribe send an email to discuss+unsubscribe@mdocml.bsd.lv

Re: uname() in mdoc_validate.c on Solaris

[Kristaps Dzonsons]

> uname(2) on Solaris (OpenSolaris, OpenIndiana, Solaris 11 Express)
> states:
>
> RETURN VALUES
>       Upon  successful  completion,  a   non-negative   value   is
>       returned.  Otherwise,  -1  is  returned  and errno is set to
>       indicate the error.
>
> Could we possibly change
>
>      if (uname(&utsname)) {
>
> to
>
>      if (uname(&utsname) == -1) {
>
> to make .Os without arguments work on these OSes?

Ingo,

This surprises me, as I usually do (-1 == syscall()).  And indeed, it 
seems this was lost in version 1.136 of mdoc_validate.c:

http://mdocml.bsd.lv/cgi-bin/cvsweb/mdoc_validate.c.diff?r1=1.135&r2=1.136&cvsroot=mdocml&f=h

Is there a technical reason for this (according to the CVS log, I was 
merging your changes)?  If not, I'll revert to the specific check for -1.

Yuri, thanks for the note!

Thanks,

Kristaps
--
 To unsubscribe send an email to discuss+unsubscribe@mdocml.bsd.lv

Re: uname() in mdoc_validate.c on Solaris

[Ingo Schwarze]

Hi Kristaps, hi Yuri,

Kristaps Dzonsons wrote on Thu, Mar 31, 2011 at 12:23:21PM +0200:
> Yuri Pankov wrote:

>> uname(2) on Solaris (OpenSolaris, OpenIndiana, Solaris 11 Express)
>> states:
>>
>> RETURN VALUES
>>      Upon  successful  completion,  a   non-negative   value   is
>>      returned.  Otherwise,  -1  is  returned  and errno is set to
>>      indicate the error.

Hm, indeed, that is not just Solaris, but POSIX, see here:

 http://pubs.opengroup.org/onlinepubs/9699919799/functions/uname.html

> This surprises me, as I usually do (-1 == syscall()).  And indeed,
> it seems this was lost in version 1.136 of mdoc_validate.c:
> 
> http://mdocml.bsd.lv/cgi-bin/cvsweb/mdoc_validate.c.diff?r1=1.135&r2=1.136&cvsroot=mdocml&f=h
> 
> Is there a technical reason for this (according to the CVS log, I
> was merging your changes)?  If not, I'll revert to the specific
> check for -1.

Actually, i did the change on purpose given the text in the
OpenBSD uname(3) manual:

  RETURN VALUES
     If uname() is successful, 0 is returned; otherwise, -1 is returned and
     errno is set appropriately.

Hence, i regarded requiring zero as safer.

However, in view of the POSIX text, my change is clearly wrong.

>> Could we possibly change
>>
>>     if (uname(&utsname)) {
>>
>> to
>>
>>     if (uname(&utsname) == -1) {
>>
>> to make .Os without arguments work on these OSes?

Yes.
Kristaps, please go ahead.

> Yuri, thanks for the note!

Indeed, and sorry for inadvertently introducing a POSIX violation.

Yours,
  Ingo
--
 To unsubscribe send an email to discuss+unsubscribe@mdocml.bsd.lv