home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
kermit.columbia.edu
/
kermit.columbia.edu.tar
/
kermit.columbia.edu
/
newsgroups
/
misc.20020314-20021006
/
000141_fdc@columbia.edu_Mon Jun 10 10:12:21 EDT 2002.msg
< prev
next >
Wrap
Text File
|
2002-10-06
|
3KB
|
70 lines
Article: 13440 of comp.protocols.kermit.misc
Path: newsmaster.cc.columbia.edu!news.columbia.edu!news-not-for-mail
From: fdc@columbia.edu (Frank da Cruz)
Newsgroups: comp.unix.solaris,comp.protocols.kermit.misc
Subject: Re: C-Kermit 8.0 for Solaris 9?
Date: 10 Jun 2002 10:12:10 -0400
Organization: Columbia University
Lines: 53
Message-ID: <ae2c3q$2r$1@watsol.cc.columbia.edu>
References: <ae08qj$iq4$1@watsol.cc.columbia.edu> <ae0e9k$1im$1@new-usenet.uk.sun.com> <ae0jqb$ptp$1@watsol.cc.columbia.edu> <ae0o8r$4at$1@new-usenet.uk.sun.com>
NNTP-Posting-Host: watsol.cc.columbia.edu
X-Trace: newsmaster.cc.columbia.edu 1023718331 13574 128.59.39.139 (10 Jun 2002 14:12:11 GMT)
X-Complaints-To: postmaster@columbia.edu
NNTP-Posting-Date: 10 Jun 2002 14:12:11 GMT
Xref: newsmaster.cc.columbia.edu comp.unix.solaris:395121 comp.protocols.kermit.misc:13440
In article <ae0o8r$4at$1@new-usenet.uk.sun.com>,
Andrew Gabriel <andrew@cucumber.demon.co.uk> wrote:
:
: OK, I'll stop speaking for other commercial UNIXs, but Solaris
: at least guarantees this. If you find some case where it fails,
: you should raise a bug on it.
:
Yes, but we don't know yet whether it works because we have spent
more time discussing why it should not be tested than it would have
taken to test it. Trusting a new operating system to be free of
bugs -- or even incompatibilities with previous releases -- is an
act of blind faith. The fact is, nobody can possibly know whether
a complex application like C-Kermit (over 280,000 lines of source at
last count), with its fingers in every nook and cranny of the file
system, network code, serial device code, and who knows what else,
will build and work in Solaris 9 without trying it.
In my experience with Unix, every new release of every Unix variety
(a) capriciously moves header files around and creates conflicts
among them that did not exist before; (b) messes with the UUCP
lockfile conventions; (c) changes how the serial driver works
(usually with respect to ill-defined Unix concepts such as hardware
flow control or high serial speeds), (d) changes the data type of
some critical quantity (e.g. from short to int, int to long), and
on and on.
Even if Solaris were perfect in preserving backwards compatibility
of binaries, we still want to make sure the application compiles
and links; otherwise if we need to make changes and old Solaris
versions are no longer available to build on, we're stuck.
Even if the Sun development environment were perfect in preserving
backwards compatibility of source code, we'd still have gcc to
worry about, thanks to Sun's enlightened policy of not including
developer tools with Solaris.
Even if all that was guaranteed worry-free, why make a special
case of Solaris when we must make new builds for every new release
of every other operating system because, in general, they can't be
trusted to preserve backwards compatibility?
So again: if somebody would please take the trouble to try building
C-Kermit on Solaris 9 with cc and/or gcc and give it a run-through,
I'd appreciate it:
ftp://kermit.columbia.edu/kermit/archives/ckx201.tar.gz
Alternatively, if I can get a guest ID on a Solaris 9 system, I'll
do the testing and any needed adaptation myself.
Thanks.
- Frank