Re: corrupt cygnus Perl ports

Re: corrupt cygnus Perl ports

[Charles S. Wilson]

Dana -
(copied to the cygwin list, because the info is probably of general interest.
Hope that's okay)

> Managed to get a valid file and set it up.

Good. It wasn't supposed to be that hard... :-)

> Running perl -MCPAN -eshell
> is not going smooth. IO::Socket is having trouble, libnet is hanging in the
> middle.

Okay, I think I know what this is. If you check the documentation (the README is
available from http://cygutils.netpedia.net/perl/README.cygwin32-v1.4.1 ) and
search for io_sock.t, you'll find that there was a problem with it. We hacked the
test so that the build would go smoothly, since without this hack, the make test
/ perl harness *hangs* rather than reporting a failure. Unfortunately, this means
that there is no warning to the unsuspecting user that io_sock doesn't work
properly, unless you read the documentation pretty thoroughly.

The problem is that "fork"s don't work with dlls in cygwin. Yet. See the recent
thread "B20.1: Fork bug (Win95)" and an older thread "fork + dlls". Hopefully
this bug, which is a cygwin problem and not a perl problem, will be fixed in B21.

>  ReadLine gets an uninitialized value error, so does Ext_MM
> (makemaker?)
>

I don't understand this. But that's just because I'm (a) not a programmer, and
(b) a verifiable, card-carrying Perl newbie. How I ended up "maintaining" the
cygwin port of perl, I'll never figure out. I guess it's just because I'm good at
collecting other people's work and packaging it up nicely. :-)

--Chuck



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

Re: corrupt cygnus Perl ports

[Charles S. Wilson]

Dana -
(copied to the cygwin list, because the info is probably of general interest.
Hope that's okay)

> Managed to get a valid file and set it up.

Good. It wasn't supposed to be that hard... :-)

> Running perl -MCPAN -eshell
> is not going smooth. IO::Socket is having trouble, libnet is hanging in the
> middle.

Okay, I think I know what this is. If you check the documentation (the README is
available from http://cygutils.netpedia.net/perl/README.cygwin32-v1.4.1 ) and
search for io_sock.t, you'll find that there was a problem with it. We hacked the
test so that the build would go smoothly, since without this hack, the make test
/ perl harness *hangs* rather than reporting a failure. Unfortunately, this means
that there is no warning to the unsuspecting user that io_sock doesn't work
properly, unless you read the documentation pretty thoroughly.

The problem is that "fork"s don't work with dlls in cygwin. Yet. See the recent
thread "B20.1: Fork bug (Win95)" and an older thread "fork + dlls". Hopefully
this bug, which is a cygwin problem and not a perl problem, will be fixed in B21.

>  ReadLine gets an uninitialized value error, so does Ext_MM
> (makemaker?)
>

I don't understand this. But that's just because I'm (a) not a programmer, and
(b) a verifiable, card-carrying Perl newbie. How I ended up "maintaining" the
cygwin port of perl, I'll never figure out. I guess it's just because I'm good at
collecting other people's work and packaging it up nicely. :-)

--Chuck



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