|
Boost : |
From: Beman Dawes (bdawes_at_[hidden])
Date: 2001-10-02 18:51:16
At 05:35 PM 10/2/2001, David Abrahams wrote:
>Anyway, I'd like to discuss with the group whether
backwards-compatibility
>is paramount. Should we maintain the status quo, crufty-up the design and
>the documentation to maintain backward-compatibility, provide a
transition
>path using #ifdefs, or something else?
If it is a minor issue, like someone coming up with a better function or
class name, we should try to maintain backward-compatibility.
But if a design is just plain wrong, or a markedly better design comes
along, or for other major issues, I don't think we should be slaves to
backward-compatibility.
A somewhat painful (for those of us on the standards committee) example:
suppose vector<bool> was in a Boost library rather than the Standard
Library. We would (or should) remove it. Not perpetuate it.
If easy to do without messing up other code, #ifdefs might be
acceptable. But remember that older versions of Boost are still available;
anyone seriously inconvenienced by removal of some component can always
fall back on a prior version.
All just my opinion, of course.
--Beman
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk