[Guido]
Is anybody maintaining the project files for Visual Studio 7.1?
[Tim]
Martin, Thomas and I all fiddle it when we notice it needs something. I view Martin as being the primary owner.
In recent checkins a new file was added (pystrtod.c) that isn't included in the project file yet.
Fixed now. I was running on Win98SE when that first got checked in, so fiddled the VC6 project files (in PC/VC6). Didn't have VC7.1 on that box.
Thanks!
I hesitate to make the checkin myself because I honestly don't understand which files are affected...
Just "the usual": used the GUI to add the new file to the pythoncore project.
Something I did to the project in response to some insistent messages at startup seems to have touched every single .proj file, which is one reason why I was hesitant to check those changes in.
test_anydbm fails, I think because of some problem with Sleepycat bsddb and/or our wrappers.
test_tcl doesn't get off the ground because it can't find tk84.dll.
Known problem, goes away when you install it.
All other "-u network,largefile" tests passed on WinXP, in both release and debug builds (although I haven't downloaded the codec test files, so those tests get skipped).
test_asynchat fails if you run it a *second* time, from .pyc/.pyo file instead of from scratch. This is a known problem with marshal not knowing how to deal with a float infinity (don't ask).
So is someone going to fix it?
test_zlib is much faster than it used to be <wink>.
There are a few other tests that are very slow, I believe test_thread is one. --Guido van Rossum (home page: http://www.python.org/~guido/)
Guido van Rossum wrote:
test_asynchat fails if you run it a *second* time, from .pyc/.pyo file instead of from scratch. This is a known problem with marshal not knowing how to deal with a float infinity (don't ask).
So is someone going to fix it?
Yes. I'll back-out #957240. Regards, Martin
[Tim]
test_asynchat fails if you run it a *second* time, from .pyc/.pyo file instead of from scratch. This is a known problem with marshal not knowing how to deal with a float infinity (don't ask).
[Guido]
So is someone going to fix it?
[Martin]
Yes. I'll back-out #957240.
Surely there was a less drastic way to fix that? sys.maxint is plenty for an "infinite" timeout... --Guido van Rossum (home page: http://www.python.org/~guido/)
On Jun 13, 2004, at 20:57, Guido van Rossum wrote:
[Tim]
test_zlib is much faster than it used to be <wink>.
There are a few other tests that are very slow, I believe test_thread is one.
test_thread is slow because it runs two tests who each theoretically take 10 seconds (randomly calculated based on the number of threads being run for the test; ``random.random() * numtasks``). We could either rewrite the test (test_dummy_thread was originally designed to replace it, but you didn't like the idea since test_thread had been around for so long and thus was stable) or lower the number of threads being run at a time to have it go faster. -Brett
participants (4)
-
"Martin v. Löwis" -
Brett -
Guido van Rossum -
Josiah Carlson