home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Usenet 1994 January
/
usenetsourcesnewsgroupsinfomagicjanuary1994.iso
/
sources
/
std_unix
/
v21
/
025
< prev
next >
Wrap
Internet Message Format
|
1990-12-05
|
1KB
From jsq Tue Aug 14 21:10:47 1990
Received: by uunet.uu.net (5.61/1.14)
id AA13384; Tue, 14 Aug 90 21:10:47 -0400
From: Donn Terry <donn@hpfcrn.fc.hp.com>
Newsgroups: comp.std.unix
Subject: Re: is struct utimbuf in the standard sys/types.h?
Message-Id: <11027@cs.utexas.edu>
Sender: fletcher@cs.utexas.edu
Reply-To: std-unix@uunet.uu.net
Date: 8 Aug 90 22:59:59 GMT
Apparently-To: std-unix-archive
From: Donn Terry <donn@hpfcrn.fc.hp.com>
(more for "is struct utimbuf".
>I am having some difficulty following the above. How can a portable
>application do anything to vendor-defined fields? Isn't the
>application non-portable as soon as it does anything (read or write)
>to a vendor-defined field?
>Is this explained by "strictly conforming" vs. "conforming"?
The point is that a portable application CAN'T know the names of the
fields (otherwise it's not portable). Given that, how does it initialize
them. It can't. Given that it can't, introducing extensions to structures
that are not initialized by the OS, and which don't have an "enable feature"
flag (as do some of the I/O related stuff with flag words) is a bad idea.
(The "enable feature" flags get you out because setting bits other than
those defined by the standard gets you into the "unspecified" space, thus
the portable application cannot set those bits.)
Donn Terry
Same old disclaimer, again.
Volume-Number: Volume 21, Number 25