On 1/14/15 7:03 AM, Gilles wrote:
> On Tue, 13 Jan 2015 20:25:53 -0700, Phil Steitz wrote:
>> On 1/13/15 6:24 PM, Gilles wrote:
>>> Hello.
>>>
>>> Now that we can run the code located in the "src/userguide"
>>> directory, I'd propose that some examples be written that
>>> produce various reports for inclusion in the user guide.
>>>
>>> A good starting point would be to move
>>>   o.a.c.m.util.FastMathTestPerformance
>>> from the "src/test" directory to "src/userguide".
>>>
>>> Rationale: it's not really a test but actually a benchmark
>>> that's only useful through _looking_ at the output.
>>
>> Why does the report itself belong in the user guide?  Does it help
>> users make better decisions about how to use [math]?  That is what
>> the User Guide is for - to help people find things and learn how to
>> use the software.  If the performance tests help make decisions,
>> then fine; if not they should not be reproduced in the guide.  Or
>> are you just suggesting we move the code there so it is easier to
>> execute?
>
> A table that quantitatively backs the claim that FastMath is a
> faster and more accurate replacement for Math is (IM not so HO)
> totally relevant to help users decide whether they should find
> and replace every occurrence in their applications.[1]
>
> That said, if you feel strongly that the user guide is not the
> best place for something like a "Commons Math Facts" chapter,
> we could create another document.

Probably belongs somewhere else on the web site.  Also will need to
be updated periodically.  I guess that is your point below.
>
> The second reason is practical, indeed: it will be easier to
> write and run something there that will produce an appropriately
> formatted[2] output without introducing spurious dependencies in
> the "test" part of the repository.

Fine by me. 

Phil
>
>
> Gilles
>
> [1] I know a distinctive case where users passed on it because
>     some unit tests failed as a result of the change, and no
>     advantage was prominent enough to warrant investigation.
> [2] Suggestions welcome.
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org
> For additional commands, e-mail: dev-h...@commons.apache.org
>
>


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org
For additional commands, e-mail: dev-h...@commons.apache.org

Reply via email to