From: Beman Dawes (bdawes_at_[hidden])
Date: 2005-01-05 08:31:16
At 12:13 AM 1/5/2005, Larry Evans wrote:
>On 01/04/2005 06:54 PM, Beman Dawes wrote:
>> Boost libraries should continue to use other Boost libraries where
>> useful, IMO. Excess coupling only becomes "excess" when it serves
>> useful purpose.
>So if variant or tuple can be implemented without coupling to other
>Boost libraries and with no disadvantage w.r.t. the current Boost
>variant and tuple, then the coupling to other Boost libraries in the
>current Boost variant and tuple is "excessive". Since there does exist
>other variant (and tuple) implementations without this coupling, the
>only remaining question is "is there any disadvantage w.r.t. the
>current Boost variant and tuple?". Agreed?
That's really a decision for the developers of those libraries. They are in
the best position to judge the impact pro and con of removing the coupling.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk