On Sat, 17 Sep 2016 10:53:42 +0200, Emmanuel Bourg wrote:
Le 17/09/2016 à 04:26, Gilles a écrit :

Sorry, but none of them warrant a -1.

You are free to completely ignore my feedback as you currently do,

Please don't pretend that I do that.
Ignore means not replying.
I never ignored any feedback.

Is it OK to not agree with your feeling that the code is not
ready for a release?

you can't claim that my concerns doesn't warrant a -1. The API is
immature and should not be released as 1.0.

My impression is that you use "immature" and "not ready
for release" in a self-referential way.

The API is
 * UniformRandomProvider
    Interface with 10 methods named according to usage in
    the Java world ("java.util.Random").
 * RandomSource
    Factory class with 9 public methods.

Everything is fully documented design-wise and usage-wise.

Usage is simple.
Contributing is simple.

You should not tag this code "immature" because you have
the impression (there is nothing more) that you can make
something better (but didn't do it for lack of time).

And if you are not listening
don't complain that you are alone.

It's a pity that I could not convince you with a thoroughly
argued email, with repeats of what is already in the docs,
and that you seem to intentionally ignore by summing it all
up with the term "immature".

Did you even use this API?

The ML archives are full of questions of mine that could have
elicited valuable feedback.
You did not provide any; it's then a not totally unexpected
consequence that the development departed from what would
have perhaps looked more natural to you.

It does not mean that it is bad, and should be thrown to
the bin only because it feels weird to a newcomer (you).

Your behaviour is in blatant contradiction with the Apache
motto, and this project's chair mantra (RERO), which is
truly ignored by the facts.

I'm all ears for setting up plans for further breaking up
Commons Math into manageable components, such as "rng-utils"
(where you can experiment with all sorts of syntactic sugar
if the core API bothers you so much).

But these ideas have been up for months, too, and not a
single PMC member did anything constructive about it, in
one way[1] or another[2]; so that the report to the board
had to mention that the CM problem has not be dealt with.

There is more serious work to do[3] rather than nit-picking
about Commons RNG.[4]


[1] Work with me towards modularity.
[2] Prepare a new release of the whole of CM, which implies
    resolving the backlog of issues).
[3] If you see what I mean...
[4] Good or bad, only time will tell.

Emmanuel Bourg

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

Reply via email to