On Sat, 20 Jul 2013 16:41:00 -0700, Ajo Fod wrote:
Ok let's approach this one step at a time starting with improper
integration because tests for AQ depend on it.
What are the modifications to the class Infinite in Math-995 that
are:
1: necessary
2: desirable
for acceptance to commons?
what bits of code can be reused from existing ones in commons? Mind
you all
the code does is map an arbitrary function in the range [-Inf, Inf]
to
[-1,1]
OK, let's reboot! :-)
First off, we should never hijack threads (see subject line of this
one); so,
please start a new thread for requesting everyone's opinion on how to
break
the functionality of your patch into its various constituents (so that
they
become independant, manageable, reusable components).
Each of these concepts should then have their own thread for discussing
the
details of implementation.
Regards,
Gilles
On Sat, Jul 20, 2013 at 2:28 PM, Gilles
<gil...@harfang.homelinux.org>wrote:
On Sat, 20 Jul 2013 06:45:10 -0700, Ajo Fod wrote:
[...]
If you want to pose what I and the community see as a respectable
objection
to including a commit, it needs to be in the form of :
1. alternative code that demonstrates a superior way of doing
improper
integration. With unit tests that show how it is faster or more
accurate.
2. OR failures of the the submitted patch in a JUnit test.
Otherwise, I maintain that the hurdle of convincing you of the
correctness/optimality of a solution is way too high for anyone's
good ...
including yours.
I admire your insistance. :-)
So, I'll try once more to get "my" point through.
A good library must avoid code duplication.
Although
* change of variables,
* improper integrals,
* adaptive algorithms
are all useful features for use in numerical estimation of
integrals, I
think that CM must not include a code like your patch, because it
provides
all of them bundled in a way that
1. does not reuse the existing code, and
2. does not allow to be reused (within CM or by CM users).
I stated that much from the outset; and that we could work
_together_ to
fulfill this requirement (and others along the way).
Once the "ability to do improper integration" is in CM, you can
always
mend/improve the patch. Its not like you've never integrated a
class or
moved it around c.f MATH-827.
Who is "you"?
That's the problem.
Had you taken the route which I suggested, most roadblocks would
have
disappeared "automagically".[1]
Regards,
Gilles
[1] Phil's too, probably, because code reuse places a big part of
the
burden (e.g. of numerical analysis in this instance) on the
reused
building blocks!
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org
For additional commands, e-mail: dev-h...@commons.apache.org
