[boost] Re: Numeric Conversion Formal Review

From: Fernando Cacciola (fernando_cacciola_at_[hidden])
Date: 2003-12-15 17:20:41

Paul A. Bristow wrote:
>> -----Original Message-----
>> From: boost-bounces_at_[hidden]
>> [mailto:boost-bounces_at_[hidden]]On Behalf Of Thorsten Ottosen
>> Sent: Monday, December 08, 2003 12:05 PM
>> To: boost_at_[hidden]
>> Subject: [boost] Numeric Conversion Formal Review
>>
>>   a.. What is your evaluation of the design?
> Looks OK, if complex.
>
> I only worry about the implications of C99 functions being included
> in the C++ Standard library (as seems very likely). This will provide
> some useful, but potentially overlapping functions.
>
AFAICT, there won't be any _overlap_, just __additional_ float 2 int
rounding functions that can be used with the Float2Int policy.

> Do we need to worry about this?  If it is now know to be accepted, it
needs at
> least to be referenced.
>
>>   b.. What is your evaluation of the implementation?
>
> Not tried to reproduce the tests.
>
>>   c.. What is your evaluation of the documentation?
>
> OK, but could be improved.  I note:
>
> 1  A fair number of spelling mistakes.  (I offer to proof re-read
> after Guillaume Melquiond's substantive comments are done).
>
Opps.. OK, thanks for the proof reading.

> 2  Lacks a novices guide to using it in practice.  I believe this is
> really necessary to ensure it gets used in practice.
OK

> 3  Lacks (over)-commenting in examples.
OK

> 4  Copyright info is outdated (last revision was Jan 2003?)
> (also Copyright outdated in code).

Ouch... right, way outdated.

> 5  I suspect more references would be helpful (C99?)
OK

> 6  Most important, it lacks some typical real-life use examples /
> scenarios.
>
OK

> (and wouldn't it be nice we could arrange boost.css to provide colour
> for code sections?
> Green for comments are especially helpful.  Can any html whizzes
> suggest how to do this easily, and portably?
>
OK.

>   d.. What is your evaluation of the potential usefulness of the
>> library?
>
> Very useful.
>>
>>   e.. Did you try to use the library?  With what compiler?  Did you
>> have any problems?
>
> No.  But the tests looked plausible.
>
>>   f.. How much effort did you put into your evaluation? A glance? A
>> quick reading? In-depth study?
>
> About 1 hour reading documentation and glancing at code.
>>
>>   g.. Are you knowledgeable about the problem domain?
>
> Not anywhere near as knowledgable as other erudite reviewers, but
> have read the Goldberg article "What everyone should know about
> floating point".  And Khan's "what you can learn about FP in 30
> minutes", so not totally novice.
>>
>>   h.. Do you think the library should be accepted as a Boost library?
>
> Yes.
>
Great!

Fernando Cacciola
SciSoft