On Jun 25, 2012, at 10:35 PM, David Cournapeau wrote:

> On Tue, Jun 26, 2012 at 4:10 AM, Ondřej Čertík <[email protected]> 
> wrote:
> 
>> 
>> My understanding is that Travis is simply trying to stress "We have to
>> think about the implications of our changes on existing users." and
>> also that little changes (with the best intentions!) that however mean
>> either a breakage or confusion for users (due to historical reasons)
>> should be avoided if possible. And I very strongly feel the same way.
>> And I think that most people on this list do as well.
> 
> I think Travis is more concerned about API than ABI changes (in that
> example for 1.4, the ABI breakage was caused by a change that was
> pushed by Travis IIRC).

In the present climate, I'm going to have to provide additional context to a 
comment like this.  This is not an accurate enough characterization of events.  
 I was trying to get date-time changes in, for sure.   I generally like feature 
additions to NumPy.   (Robert Kern was also involved with that effort and it 
was funded by an active user of NumPy.    I was concerned that the changes 
would break the ABI.  In fact, I expected them to --- I was not against such 
changes, even though it was a change in previously discussed policy.  We just 
needed to advertise them widely.       Other voices, prevailed, however, and 
someone else believed the changes would not break ABI compatibility.   
Unfortunately, I did not have much time to look into the matter as I was 
working full time on other things.    

If I had had my way we would have released NumPy 1.5 at the time and widely 
advertised the ABI breakage (and moved at the same time to a design that would 
have made it easier to upgrade without breaking the ABI).    I do not believe 
it would have been that big of a deal as long as we communicated correctly 
about the release.     I still don't think it's correct to be overly concerned 
about ABI breakage in a world where packages can just be re-compiled against 
the new version in a matter of minutes with one hand and with the other make 
changes to the code base that change existing code behavior.    I think the 
fact that the latter has occurred is evidence that we have to sacrifice one of 
them.   And ABI compatibility is the preferred one to sacrifice by a long 
stretch in my view. 

-Travis




_______________________________________________
NumPy-Discussion mailing list
[email protected]
http://mail.scipy.org/mailman/listinfo/numpy-discussion

Reply via email to