This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: BSD test of ABI checker code
- From: Benjamin Kosnik <bkoz at redhat dot com>
- To: libstdc++ at gcc dot gnu dot org
- Date: Tue, 20 Aug 2002 21:24:33 -0700
- Subject: Re: BSD test of ABI checker code
- Organization: Red Hat / San Francisco
- References: <200208202259.g7KMxE9K073093@latour.rsch.comm.mot.com>
Thanks Loren!
> Without the explicit `mv 3.2_symbols-2.txt 3.2_symbols.txt', I
> received no error that file, 3.2_symbols.txt, could not be opened and
> ~2000 missing symbols were detected...
Yeah. Sorry about that. (Error checking, what's that? Seems to be the
chronic issue with my scripts.)
> Of course, I moved the Linux baseline out of the way since we don't
> export wchar symbols, etc and as predicted there are some trivial size
> differences related to standard typedefs, etc.
I must admit, I'm curious as to the actual difference. Right now
incompatible symbols spew out a bunch of info, including sizes for
objects. Can I see that please?
How many symbols total are you exporting versus linux? If you
generate the list of symbols for your platform's 3.2.0 release and
compare it to head gcc, do you see the same missing symbols?
It looks like os-cpu matrix of baseline test results is necessary. Damn.
The configure stuff is organized for cpu or os, but not both.
-benjamin