Alain, this is the issue that we saw as well. open vs fopen. I hope you are able to work it out! I may have to leave certain machines on 7.2 to support the program marparse until a solution is resolved. Its easier than downgrading to 2.2.
Paul
----- Original Message ----- From: "Alain Knaff" alain.knaff@lll.lu To: "Paul Murray" pjmurray@tfb.com; "Marnus van Niekerk" m@mjvn.net Cc: zlibc@tux.org Sent: Thursday, May 29, 2003 6:17 AM Subject: Re: [zlibc] Redhat 8
On Wednesday 28 May 2003 17:50, Paul Murray wrote:
Marnus thanks for the info. I wanted to make sure I wasn't doing
something
wrong. "more" and another program "marparse" are the programs that I need to
get
working...with marparse being the most important. I sent Alain a verbose strace of marparse so Alain could see what calls were being made and perhaps see where the code is behaving differently.
Paul
I looked more closely into the issue, and noticed that glibc 2.3 has changed the way how libc internal functions call other libc internal functions. Formerly, they would pick up zlibc's replacements, but they no longer do now (many libc function set up "hidden" definitions using the libc_hidden_def macro, that can no longer be overriden by outside programs).
Thus applications that direclty use open() continue to work with zlibc, but those that use fopen64() don't, because fopen64() now always calls libc's own open64, and never zlibc's.
I've notified Ulrich Drepper and Jakub Jelinek from the glibc development team about the issue; hopefully we'll be able to work out a solution together.
In the meantime, I'd suggest you downgrade to glibc 2.2
Regards,
Alain
_______________________________________________ zlibc mailing list zlibc@tux.org http://www.tux.org/mailman/listinfo/zlibc