[Cfp-interest 1348] obsolescing log1p

[Jim Thomas]

Jens suggests that we obsolesce log1p, in favor of logp1, and add a statement about why the name changed. This will be on the agenda for the July CFP teleconference. Below is an email discussion between Jens and myself (blue text) about it.

- Jim Thomas

>> I think we should give the users of these functions some guidance. I
>> would prefer to do just the usual stuff in the description section
>> and then add a note, that the systematic naming approach of C is
>> now the `p1` suffix, and that the `log1p` interface might be
>> removed at some point. This also tells the worried user that, yes,
>> in fact these functions are the same, but tells them a bit more
>> about the reasons and where to go.  
> 
> I like your idea of pointing out why the new name was introduced,
> whether or not we obsolesce log1p.
> 
> There is a cost to obsolescing. If an identifier they reference is
> obsolesced, users have to deal with the risk their code will break in
> the future.

Yes, but nobody can say that we have been very agressive on that
front :)

The promis always is that such an identifier survives at least to the
next revision cycle. Any software that is maintained should be able to
deal with that.

> I’ll bring this up at the next CFP meeting. If there’s consensus I’ll
> propose the change to obsolesce log1p to WG14. Of course, you could
> propose it if you wish. 

No, no, I largely prefer that you guys get into this.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.oakapple.net/pipermail/cfp-interest/attachments/20190713/af0b10ca/attachment.html