Hello,

In message "Re: [ ruby-Bugs-1200 ] Segmentation Fault with Sockets and Threads"
    on Dec.14,2004 00:56:21, <matz / ruby-lang.org> wrote:
| |Initial Comment:
| |With ruby 1.8.2 (2004-11-06) [i386-mswin32], the following code segfaults after about 12 seconds.  Removing socket.close stops the fault.
| 
| Could someone check these bugs (#1200, #1190, #1189, #1188) against
| the latest 1.8.2 snapshot?

#1188:
With 1.8.2, it causes IOError on mswin32/mingw32.
I think there are some problems in IO#reopen with stdin/stdout/stderr,
not only mswin32/mingw32. (I heared that NetBSD has another problem
on this test).

#1189:
1.8.2 doesn't have this test. So, no problem :)
This is well-known "win32 console/pipe + select" problem.

#1190:
CVS HEAD only.
CVS HEAD uses fake socketpair as bidirectional pipe.
But it causes Errno::EINVAL if the socket is closed by child
process.

#1200:
I have reproduced the problem on both 1.8.2 and CVS HEAD.
I'm checking it now.


Regards,
-- 
U.Nakamura <usa / garbagecollect.jp>
# sorry for my poor English... orz