[Numpy-discussion] Release blockers for 1.4.0 ?
Mon Dec 7 22:15:30 CST 2009
Charles R Harris wrote:
> On Mon, Dec 7, 2009 at 10:31 AM, David Cournapeau <firstname.lastname@example.org
> <mailto:email@example.com>> wrote:
> On Tue, Dec 8, 2009 at 1:48 AM, Charles R Harris
> <firstname.lastname@example.org <mailto:email@example.com>> wrote:
> > On Mon, Dec 7, 2009 at 8:24 AM, David Cournapeau
> <firstname.lastname@example.org <mailto:email@example.com>> wrote:
> >> Hi,
> >> There are a few issues which have been found on numpy 1.4.0,
> which worry
> >> me:
> >> # 1317: segfaults for integer division overflow
> >> # 1318: all FPU exceptions ignored by default
> >> #1318 worries me the most: I think it is a pretty serious
> >> since things like this go unnoticed:
> >> x = np.array([1, 2, 3, 4]) / 0 # x is an array of 0, no warning
> > Hasn't that always been the case? Unless we have a way to raise
> > from ufuncs I don't know what else we can do.
> No, it is a consequence of errors being set to ignored in numpy.ma
> So the fix is easy - but then it shows many (> 500) invalid values,
> etc... related to wrong fpu handling (most of them are limited to the
> new polynomial code, though).
> Umm, no. Just four, and easily fixed as I explicitly relied on the
Yup, my last sentence was not really clear: I just meant that most of
the warnings were printed during polynomial tests.
> After the fix and seterror(all='raise'):
> FAILED (KNOWNFAIL=4, SKIP=11, errors=47, failures=1)
47 errors is quite a bit.
So what about the following for 1.4.0:
- we just remove the seterr set to ignore in numpy.ma
- we deal with the special case of MIN_INT/-1 to avoid the SIGFPE crash
- we deal with the spurious warnings during tests in 1.5.0 ?
More information about the NumPy-Discussion