Re: proposal: minscale, rtrim, btrim functions for numeric

út 10. 12. 2019 v 0:03 odesílatel Karl O. Pinc <kop(at)meme(dot)com> napsal:

> On Mon, 9 Dec 2019 21:04:21 +0100
> Pavel Stehule <pavel(dot)stehule(at)gmail(dot)com> wrote:
>
> > I fixed almost all mentioned issues (that I understand)
>
> If you don't understand you might ask, or at least say.
> That way I know you've noticed my remarks and I don't
> have to repeat them.
>
> I have 2 remaining suggestions.
>
> 1) As previously suggested: Consider moving
> all the code you added to numeric.c to right after
> the scale() related code. This is equivalent to
> what was done in pg_proc.dat and regression tests
> where all the scale related stuff is in one
> place in the file.
>
> 2) Now that the function is called min_scale()
> it might be nice if your "minscale" variable
> in numeric.c was named "min_scale".
>
> I don't feel particularly strongly about either
> of the above but think them a slight improvement.
>

done

> I also wonder whether all the trim_scale() tests
> are now necessary, but not enough to make any suggestions.
> Especially because, well, tests are good.
>

I don't think so tests should be minimalistic - there can be some
redundancy to coverage some less probable size effects of some future
changes. More - there is a small symmetry with min_scale tests - and third
argument - some times I use tests (result part) as "documentation". But I
have not any problem to reduce tests if there will be requirement to do it.

Regards

Pavel

> Regards,
>
> Karl <kop(at)meme(dot)com>
> Free Software: "You don't pay back, you pay forward."
> -- Robert A. Heinlein
>