that old GNU/Linux argument

Les Mikesell lesmikesell at
Fri Jul 25 19:19:29 UTC 2008

Alan Cox wrote:
>> Yes, there was some version of the Posix standard in that time frame. 
>> It was just incomplete and described some mythical system that matched 
>> no existing BSD or SysV flavor, so it was mostly ignored.  Sort of like 
> Not really the case.
> POSIX described a set of behaviours that were Unixlike and could be
> relied upon.

Except that when they were written, no system exactly matched what they 
specified so you couldn't rely on them to work although they might have 
been useful to point fingers at the non-complying implementations.

> What it covered was far less than Unix and it took great
> care to indicate what was not to be relied upon and in time SuS and
> similar specs expanded on this by introducing new functions where
> commonality was needed  and the existing interfaces were deficient -
> obvious examples include termios and sigaction.
> POSIX was and is very important but the basic core of posix isn't about
> 'being Unix' it is about fundamental things like
> write/lseek/open/mkdir ...

It's all somewhat academic until you have enough that a program actually 
works on more than one platform.  I think the 1995 version started to 
get there.  I'm not sure which one Windows NT used as it's checklist 
item to claim conformance.

   Les Mikesell
    lesmikesell at

More information about the fedora-list mailing list