[Python-bugs-list] [ python-Bugs-233050 ] Python-2.1a2 tests failed on NetBSD 1.5 i386

[noreply@sourceforge.net]

Bugs item #233050, was updated on 2001-02-18 23:12
You can respond by visiting: 
http://sourceforge.net/tracker/?func=detail&atid=105470&aid=233050&group_id=5470

Category: Build
Group: Platform-specific
>Status: Closed
Priority: 3
Submitted By: Anders Andersen (aandersen)
Assigned to: Guido van Rossum (gvanrossum)
Summary: Python-2.1a2 tests failed on NetBSD 1.5 i386

Initial Comment:
Experiences from compiling and installing Python-2.1a2 on a NetBSD 1.5
i386 machine with the pth 1.3.7 thread implementation (thread library
installed as a NetBSD package and located in the /usr/pkg directory):

Configure done with the following command:

  CFLAGS=`pthread-config --cflags` \
  LDFLAGS=`pthread-config --ldflags` \
  LIBS=`pthread-config --libs` \
  ./configure 

Added thread information to CFLAGS and LDFLAGS in Makefile:

  CFLAGS=   $(OPT) -I/usr/pkg/include -I. -I$(srcdir)/Include $(DEFS)
  LDFLAGS=  -L/usr/pkg/lib

Build Python-2.1a2 with the `make' command.  Run tests (make test)
with the following results:

  Skipped: al, bsddb, cd, cl, dl, gdbm, gl, imgfile, linuxaudiodev,
  minidom, nis, pyexpat, sax, sunaudiodev, sundry, winreg, winsound

  Failed: fork1, largefile, mailbox, ucn

I noticed the following strange problems: `make test' hangs forever on
test_thread, but test_thread works OK done separately.  test_pty hangs
forever done separately, but works OK during `make test'.

Results from failed tests:

  $ ./python Lib/test/test_fork1.py
  Traceback (most recent call last):
    File "Lib/test/test_fork1.py", line 72, in ?
      main()
    File "Lib/test/test_fork1.py", line 47, in main
      verify(a == range(NUM_THREADS))
    File "./Lib/test/test_support.py", line 81, in verify
      raise TestFailed(reason)
  test_support.TestFailed: test failed

  $ ./python Lib/test/test_largefile.py
  Traceback (most recent call last):
    File "Lib/test/test_largefile.py", line 18, in ?
      f.seek(2147483649L)
  IOError: [Errno 22] Invalid argument

  $ ./python Lib/test/test_mailbox.py  
  Traceback (most recent call last):
    File "Lib/test/test_mailbox.py", line 23, in ?
      try: os.rmdir(newdir)
  OSError: [Errno 20] Not a directory: '@test/new'

  $ ./python Lib/test/test_mailbox.py 
  ...
  Testing unicode character name expansion strict error handling....
  Traceback (most recent call last):
    File "Lib/test/test_ucn.py", line 90, in ?
      raise AssertionError, "failed to raise an exception when given
	a bogus character name"
  AssertionError: failed to raise an exception when given a bogus
	character name


----------------------------------------------------------------------

>Comment By: Guido van Rossum (gvanrossum)
Date: 2001-04-10 13:39

Message:
Logged In: YES 
user_id=6380

I just fixed the test_largefile failure -- on some (64bit?)
systems f.seek() with a value > 2**32 raises IOError, not
OverflowError.

The failure in test_largefile caused it to leave behind a
file @test which prevented test_mailbox from doing
mkdir('@test'). I also fixed this today.

I can't help you with the fork1 and ucn failures or with the
threading; AFAIC, pth is not stable enough to worry about. 

Hence, I'm closing the bug report.  If you find the causes
for the problems and can provide patches relative to current
CVS, please open a new patch entry.

----------------------------------------------------------------------

Comment By: Guido van Rossum (gvanrossum)
Date: 2001-03-01 21:47

Message:
Logged In: YES 
user_id=6380

Thanks for the reports.  Can't do much about most of these.

The test_mailbox.py failure (the real one) should be fixed
in 2.1b1,
I just checked in a fix that makes it test for the right
exception.

----------------------------------------------------------------------

Comment By: Anders Andersen (aandersen)
Date: 2001-02-18 23:18

Message:
The last test was test_ucn and not test_mailbox.  Sorry
about the cut and paste error.


----------------------------------------------------------------------

You can respond by visiting: 
http://sourceforge.net/tracker/?func=detail&atid=105470&aid=233050&group_id=5470