Hi, Thanks for your feedback, I will take it into account to improve the documentation. Actually I wasn't sure how long and detailed it was expected to be, so I preferred to do something short at first. Also for the description of error handling, I didn't want to end up with something too repetitive regarding what is already described in the Boost.Math error handling documentation. But indeed, a short and concise enough demo could be of help. About the "known in advance" expression, I myself hesitated a lot. The choice depends on who we prefer to talk to : the mathematician (for whom "fixed" and "known in advance" should be more natural) or the programmer (who will find "known at compile-time" more expressive). Thanks Bruno On Feb 17, 2008 7:35 PM, Paul A Bristow <pbristow@hetp.u-net.com> wrote:
-----Original Message----- From: boost-bounces@lists.boost.org [mailto:boost-bounces@lists.boost.org] On Behalf Of Bruno Lalande Sent: 17 February 2008 17:04 To: boost@lists.boost.org Subject: Re: [boost] [Boost] Compile-time power of a run-time base
Here is a new version of the pow.hpp file with updated tests and docs. It nows handles errors using the Boost.Math overflow error policy, and uses promote_args to determine the return type.
Nice.
I've tested it on the same platforms as before, BUT it no longer compiles with gcc-2.95.4. The first reason was that policy.hpp includes <limits> that gcc 2.95 doesn't have so I replaced it by <boost/limits.hpp> and it worked, but some other Boost files include other headers that are also missing (ios for instance, included in lcast_precision.hpp). Is the support for that compiler required?
Not by me ;-) I would not spend time on this outdated compiler. Surely people who really want to use a compile tile pow will be using more modern compilers?
Docs might provide a reference or few. I found
http://www.scl.ameslab.gov/Publications/Gus/FiveMultiplications/node9.html
by a very quick Google
I'm sure we should give Knuth his due for having done everything before we even thought of it ;-)
D.E. Knuth, The Art of Computer Programming, Vol. 2: seminumerical Algorithms, 2nd ed. Addison-Wesley, Reading, MA, 1981.
I think occurrence has two cs and two r as well ;-)
Would the words 'compile-time integer integral power' ensure that this doucmentation pops up when Googling?
(Would the expression 'known at compile time' be clearer than 'known in advance'?)
Should there be a less formal description of the user interface - perhaps simplified from the two actual templates in pow.hpp? It looks more complicated than it really is in pow.hpp.
Would what I might call a 'demo' program show its use more simply for novices than the formal test ? It could show use of policies - something that may frighten some users - it did me at first ;-) It could show how especially useful is using try/catch is with policies?
Paul
--- Paul A Bristow Prizet Farmhouse, Kendal, Cumbria UK LA8 8AB +44 1539561830 & SMS, Mobile +44 7714 330204 & SMS pbristow@hetp.u-net.com
_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost