Steffen Daode Nurpmeso added the comment:
I think that will also be healed with the patch i've just posted
to #11277.
(However, though i can't ,reproduce`, it seems the
Microkernel/IOKit interaction sometimes has a problem, because
i *really* had messed up sparse files (about 1-2 percent or so)
Ned Deily added the comment:
Maybe that buildbot machine is just overloaded. If I understand the buildbot
waterfall output correctly, it looks like there are often 4 simultaneous test
runs happening on that machine, one each for 2.7, 3.1, 3.2 and 3.x.
http://www.python.org/dev/buildbot/all/w
STINNER Victor added the comment:
Timing on the x86 Tiger buildbot: "time ./python.exe -m test -v -u largefile
test_mmap" gives approx 5 minutes. The buildbot is a Mac mini, Intel Core Duo
1.8 GHz, 2 GB of memory, Mac OS X 10.4.10, HD: 232 GB (182 GB available) WDC
WD2500BEVT-00A23T0.
/usr/s
STINNER Victor added the comment:
We can use the following function to check if the filesystem does support
sparse files:
def support_sparse_file(directory):
import tempfile
with tempfile.NamedTemporaryFile(dir=directory) as tmpfile:
# Create a file with a size of 1 byte but w
Ned Deily added the comment:
That said, on my iMac (2.4 Ghz Intel Core 2 Duo):
$ time ./python -m test -v -u largefile test_mmap
[...]
--
Ran 24 tests in 87.673s
OK
1 test OK.
real1m27.875s
user0m0.186s
sys 0m8.6
Ned Deily added the comment:
Note that the various HFS variants, the default file system types on OS X, do
not support sparse files at all. So any tests of large files require at some
point that the system writes out all the data in the file.
http://developer.apple.com/library/mac/#documentat
STINNER Victor added the comment:
Hum, it does still fail with a timeout of 1 hour:
---
[ 41/354] test_mmap
Timeout (1:00:00)!
Thread 0x7fff70439ca0:
File
"/Users/pythonbuildbot/buildarea/3.x.hansen-osx-x86/build/Lib/test/test_mmap.py",
line 685 in