This is the mail archive of the cygwin@sourceware.cygnus.com mailing list for the Cygwin project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: Problem compiling perl module Term::ReadKey under cygwin


 >
> >   This is the second complaint in two days I've gotten on this
> > cygwin-perl-/tmp issue.
> >
> This is arguably a lame excuse for your patch since as you said above,
> "...hack to work around an error in the perlbuild scripts ('\'
> shouldn't be used as a path separator)..."
> 
> Are we then supposed to change the OS and tools in order to excuse
> ourselves from fixing the bugs in our applications?
> Try this analogous scenario. Pragrammer to Microsoft Engineer: "I keep
> getting null pointer exceptions in my program, you should change Windows
> so that references to uninitialized data will now return the constant
> value 0xDEADBEEF instead of causing an exception."
> Let's stick to a convention for awhile, please!  The current one has my
> vote.
> 

Steve -
  That was not my patch. @#$!@#$. I happen to think that other guy, the
one who initially presented the patch to allow the behavior I described,
was right. Everyone else on the planet disagrees with both him and me.
Fine. That's okay, that's what open source is all about.

Your example above is flawed, though. If MS originally threw
'0xDEADBEEF' in response to a reference to uninitialized data (stupid as
that may be), then you would expect thousands of programs to test for
'0xDEADBEEF' -- and respond appropriately, such as running atexit() or
performing cleanup before shutting down. Then, MS comes along and says,
"OK, boys, from now on, we're gonna throw an exception. Instant coredump
-- you can't cleanup before exiting."

Wouldn't it be understandable if the programmers asked for a return to
the earlier behavior? Personally, I think that the mere presence of '\'
in a path should not be enough to indicate Windowsism. But reasonable
people can disagree, and your side won. For now. <g>

In any case, I find it ironic that everyone's being so hard on that guy.
He did what we all want people to do on this list!!! He

(a) noticed a problem
(b) proposed a patch
(c) accepted criticism
(d) proposed a revised patch

It doesn't matter if the patch wasn't accepted. The fact is, on one hand
we criticize folks for whining about bugs without attempting to fix it
themselve -- we tell them to submit a patch. That does not mean we have
to accept all patches that come in -- but when one DOES come in, let's
not ridicule the submitter, OK?

FWIW, I *did* track down this error. It's actually in libiberty, not
perl, and is the result cygwin's unusual status as a unix platform
running under windows: _WIN32 is defined, so libiberty uses '\' as the
path separator. Once I figure out to whom the patch should go (since
everybody seems to maintain their own version -- gcc, cygwin, binutils,
etc) I'll submit it.

--Chuck

--
Want to unsubscribe from this list?
Send a message to cygwin-unsubscribe@sourceware.cygnus.com


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]