Re: [gawk-devel] MPFR thoughts

[arnold]

Hi Neil.

This is the tail end of a private discussion that accidentally leaked
onto the public list.

I won't go into the details here; suffice it to say that the recent
brouhaha was the straw that broke the camel's back. I've been unhappy
about the miniscule ROI of maintaining the MPFR code for a VERY long time.

This prompted last week's request to this list for a volunteer to step
up to the plate and take over responsibility for that part of the code.
(So far, no takers.)

Assuming that nobody volunteers to take over MPFR support, the current
plan is to deprecate it in an upcoming release (early summer, I hope,
but no promises) and to remove it in 2024.

You are the only one (that I remember) that has found MPFR support in
gawk to be of practical use. There may be others, and if the deprecation
causes a flood of complaints from legitimate users, I will reconsider.

Thanks,

Arnold

"Neil R. Ormos" <ormos-gnulists17@ormos.org> wrote:

> Carl Friedberg wrote:
> > [... Manuel Collado wrote:]
> >> [Eli Zaretskii wrote:]
>
> >>> ...
> >>> Gawk is a general-purpose language,
> >>> and as such, it isn't outlandish
> >>> for Gawk to provide extended-precision arithmetics.
>
> >> Please let me respectfully disagree. IMHO gawk
> >> is a kind of domain-specific language. Focused
> >> of processing textual representation of data,
> >> i.e. strings. Of course some numeric
> >> computation capability is needed, but only to
> >> some extension.
>
> >> Not being the main purpose of gawk, this
> >> capability can be provided as an extension, as
> >> already said. No need to be integrated in the
> >> gawk core.
>
> > Thank you. I agree with the removal of MPFR
> > support. I am not in any way an AWK/GAWK
> > developer, nor do I have the skills needed.
>
> I am not a member of the Gawk developers mailing list, but since thoughts 
> about removing MPFR from Gawk have been mentioned on this Bug-Gawk list, I 
> offer this user comment.
>
> I appreciate the GNU MP and MPFR support in Gawk and urge that it not be 
> removed.  I have found the MPFR feature quite useful, e.g., for 
> experimentation to explore the behavior of a calculation when spreadsheets 
> didn't work.
>
> Even if Gawk is seen by many as primarily a text processing language, Gawk 
> has proven to be amazingly versatile in my practice, providing a marvelous 
> tradeoff between functionality and simplicity.
>
> There may be well be better languages for specific problem domains, 
> particularly for production systems, but Gawk is often quite adequate for 
> proof-of-concept and other prototyping, avoiding the large up-front cost of 
> acquiring fluency in some special-purpose language, along with its required 
> tool chain and infinity of libraries and modules.  Removing the MPFR support 
> from Gawk would reduce its versatility.
>
> Finally, it would be a shame to let the undiplomatic complaints of a single 
> user serve as the catalyst for the removal of important functionality from 
> Gawk.  To the extent Gawk -M behavior is library-dependent and cannot 
> feasibly be reconciled with Gawk's non-dash-M behavior, instead of removing 
> MPFR support, it would seem sufficient to warn users that the behavior may 
> vary and that such variation will not, alone, be treated as a bug.
>
> > We owe Arnold Andy, and all the others who have
> > supported AWK over the generations a huge thank
> > you
>
> Indeed.  Thank you Arnold, Andy, Eli, and all the others.