[SciPy-dev] Matlab io bug; request for advice

Robin robince@gmail....
Thu Feb 19 03:07:58 CST 2009


On Thu, Feb 19, 2009 at 9:05 AM, Robin <robince@gmail.com> wrote:
> On Thu, Feb 19, 2009 at 9:00 AM, Nathan Bell <wnbell@gmail.com> wrote:
>> On Thu, Feb 19, 2009 at 3:22 AM, Matthew Brett <matthew.brett@gmail.com> wrote:
>>>
>>> I'm happy either way.  It's easier for me, and might be cleaner in the
>>> long run, to just fix and document, but I can also see that broken
>>> code is not a good outcome.
>>>
>>
>> I vote fix and document in 0.7.1.  I don't think a prolonged
>> transition period would help matters.
>
> As a user of the MATLAB functionality, I would vote for the warning
> message in 0.7.1 and the change in 0.8 (with an option to keep the
> current behaviour).
>
> I use MATLAB files a lot to store results/data and it would/will be a
> pain to change everywhere I load files that I've previously saved (and
> then have to keep track of which data files were saved with the old
> version and the new which is why the compatibility option is
> important). I'm sure a lot of other people are in a similar
> situation... I think we really need a warning and a grace period - I
> wouldn't expect this sort of possible breakage in a point release.
>
> Cheers
>
> Robin

PS - to be clear - (I also forgot to differentiate) - fine with the 0d
fix going in, but my above comments apply to the change in
dimensionality of a MATLAB saved/loaded 1d vector.


More information about the Scipy-dev mailing list