This is the mail archive of the libc-hacker@sourceware.cygnus.com mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

new version of new testsuite



I've put up another new version of my improved testsuite.
ftp://rabi.phys.columbia.edu/pub/zack/testsuite-0.4.tar.gz

I moved a lot of stuff around, so there are no diffs.  I'll start
providing diffs with the next one.

Highlights of this release:

last vestiges of DejaGNU removed
now tests everything that the existing test suite does
improved command line interface for 'make check'

Quickie use instructions are in the README, and preliminary docs in
testsuite.texi.  Much improved docs are coming soon.

You should avoid running the test script with bash, it is very slow.

---

There are 54 'expected failures' in the suite.  These are all real
bugs, but some of them may be hard to fix.

21 of the failures are problems with POSIX conformance of our regexp
library.  This is minor.  An additional 11 are problems with POSIX
conformance of the headers, which is important but not enough to delay
a release.

There are two failures in the localedef test.  I believe these
correspond to outstanding bug reports.  If at all possible they should
be fixed.

The test of scanf() gets one failure.  I'm not sure if the test is
correct, it's an odd case involving %*s and %n.

backtracesyms() gets dinged for not providing any useful information
about addresses in the program.  We don't have all the information
we'd like to provide, but we could at least print the code module
(executable, shared library, dynamic object) and its load address for
each trace entry; then addr2line could be extended to read this format
and produce a useful trace.

There are three failures in C89 conformance of the headers and twelve
in C9x conformance.  Some of these are bad interactions between gcc
headers and my very picky test program.  The one remaining notable
problem is that tgmath.h still has scalb() in it.

---

There are 12 'unsupported testcases'.  Two (inet-ifindex and POSIX
mqueue.h) are kernel issues.  The other ten are because lgammal,
erfl, and erfcl are not implemented on x86.

zw


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]