[Numpy-discussion] Warning on http://scipy.org/ about binary incompatibility ?

josef.pktd@gmai... josef.pktd@gmai...
Thu Jan 28 08:01:13 CST 2010


On Thu, Jan 28, 2010 at 1:39 AM, David Cournapeau <david@silveregg.co.jp> wrote:
> Charles R Harris wrote:
>>
>>
>> On Wed, Jan 27, 2010 at 6:20 PM, David Cournapeau <david@silveregg.co.jp
>> <mailto:david@silveregg.co.jp>> wrote:
>>
>>     josef.pktd@gmail.com <mailto:josef.pktd@gmail.com> wrote:
>>      > Can we/someone add a warning on the front page http://scipy.org/
>>      > (maybe under news for numpy download) about incompatibility of the
>>      > binaries on sourceforge of scipy <=0.7.1 with numpy 1.4.0 ?
>>
>>     It seems that it will be quite difficult to fix the issue without
>>     removing something (I tried to use datetime as user types, but this
>>     opened a can of worms), so I am (quite reluctantly ) coming to the
>>     conclusion we should just bite the bullet and change the ABI number (so
>>     that importing anything will fail instead of crashing randomly).
>>
>>     Something like numpy 1.4.0.1, which would just have a different ABI
>>     number than 1.4.0, without anything else.
>>
>>
>> Why do you think it would be better to make this change in 1.4 rather
>> than 1.5?
>
> Because then any extension fails to import with a clear message instead
> of crashing as it does now. It does not matter much if you know the
> crash is coming from an incompatible ABI, but it does if you don't :)

I thought we could get away with a small binary incompatibility,
without rebuilding everything. I'm using matplotlib although not
extensively and it didn't crash in a while. (I don't remember which
version of scipy I used for the last time when I had a crashing
script.)

I just tried to build h5py which does not import at all with 1.4.0,
but I only get compiler errors about using headers only with Visual
C++ and  conflicting types for 'ssize_t' .

Is there a way to find out which extensions use the binary
incompatible part. Since it took a long time to confirm the ABI
breakage, I would think it's not in a heavily used part. Although, I'm
not sure since I had to rebuild for example most scikits with 1.4
either because of the cython issue or because of this.

Personally, I would remove the ABI breakage for 1.4.1, I rather have a
working SciPy than a new "experimental" feature.

That's my opinion as a consumer of binary distributions.

Josef

>
> cheers,
>
> David
> _______________________________________________
> NumPy-Discussion mailing list
> NumPy-Discussion@scipy.org
> http://mail.scipy.org/mailman/listinfo/numpy-discussion
>


More information about the NumPy-Discussion mailing list