uname -s question

[Christopher Faylor]

On Sat, Feb 24, 2001 at 07:43:43PM -0500, Norman Vine wrote:
>Christopher Faylor writes:
>>
>>On Sat, Feb 24, 2001 at 07:21:37PM -0500, Norman Vine wrote:
>>>Christopher Faylor writes:
>>>>>Which would enable constructs like the following to work:
>>>>>
>>>>>    if sys.platform in ['cygwin', 'linux']:
>>>>>        # ...
>>>>
>>>>Can't you do something equivalent with regular expressions in python?
>>>
>>>Of course, we are doing something 'equivalent' and will 
>>contiue todo so.
>>
>>I was trying to get a handle on why this was a problem.  Judging by
>>usages in most of the configure scripts in sources.redhat.com and by
>>perl's Configure, the convention for many systems is to refer to the
>>system name with a wildcard (e.g., netbsd*), when necessary.
>>
>>I was trying to determine why this was not acceptable for Python.
>
>FYI
>Here is the original comment that led to this question.
>
>>>Looking at the GuessOS helper in the Apache source distribution I'd say
>>>that cygwin is completely non-standard in returning version information
>>>with uname -s.

I didn't see any occurrence of wildcard system matching in the GuessOS
for apache 1.3.x however the 2.0 versions of apache have wildcard tests
for the following systems:

Windows NT, Amiga, arm, Pyramid, atari, falcon, milan, MiNT, Ultrix, Delta 88K,
IRIX, Cygwin, Mingw, Interix, Uwin.

Again, I'm just trying to understand why Cygwin is a special case when
it seems that wildcard tests are normal for configuration tests.

It is not a big deal.  We've probably beaten this to death.  I was really
just curious.

cgf