On Sat, Apr 05, 2003 at 09:37:24PM +0300, Enache Adrian wrote:

FreeBSD 5.0 / user-threads (libc_r.so) in fact. According to
Jos Boumans, it works fine with FreeBSD 4.x. And probably using
the rfork-based linuxthreads or the new libthr will make it
work fine too ( I'll try the latest when I have enough time )
Should make it a TODO test be politically correct ?
That will give the impression that the failure is related to Encode
which in fact it isn't.
-close STDOUT;
+close STDOUT unless $^O eq 'freebsd';
+ # weird bug in FreeBSD5.0/user-threads (libc_r.so)
Eeugh. Because

1: From your description it's a bug in FreeBSD, isn't it? In which case it
will probably get fixed at some point. So maybe it should be a TODO.

except that

2: If it's a TODO on FreeBSD then it might pass on the other two thread
implementations you describe, and hence be a TODO test that unexpectedly
passes. (Which is also bad, IIRC)

but can we detect what sort of threads FreeBSD has from %Config::Config?

And is the only problem about this about open files and Win32? The close
isn't actually part of the Encode functionality that we're testing, is it?
So maybe it's not actually enough merit to be a TODO, and the close STDOUT
is conditional on $^O just as you suggest.

Doing that is faster than reading this message, let alone trying to make
sense of it. :-)

Nicholas Clark

PS This message supplied without warranty of any form, particularly
suitability for any purpose, including but not limited to making sense,
or furthering the development of perl. Use at your own risk. May contain
traces of nuts. Your sanity will be at risk if you fail to make payments
on debts secured against it.

Search Discussions

Discussion Posts


Related Discussions

Discussion Navigation
viewthread | post
posts ‹ prev | 3 of 3 | next ›
Discussion Overview
groupperl5-porters @
postedApr 5, '03 at 3:28p
activeApr 6, '03 at 4:06p



site design / logo © 2022 Grokbase