[Wellington-pm] Meetings

Kent Fredric kentfredric at gmail.com
Wed Aug 5 05:35:20 PDT 2015


On 5 August 2015 at 10:25, Florent Mara <florent.mara at nzregen.co.nz> wrote:

> The performance benchmarks are quite grand and since my code is mostly
> 'non-perl-magical', is it worth giving it a try and get more processing
> done out of CPU time?
>


The opinion we've been seeing based on observations in #toolchain is
largely that, at least presently, RPerl is not something you can throw
real-world code at on CPAN.

Its seems to be more tailored to writing a restricted subset of perl which
is almost C++, such that the code can be trivially converted to C++.

This means if you write code *for RPerl* that it should be portable to Perl
as-is, but the inverse is not true.

And its estimated that a potential deliverable RPerl may one day achieve is
the ability to write Perl in the RPerl subset to smooth over creation of XS
Modules, by allowing you to write them in a perl subset which can be
compiled to XS.

But outside that, we presently don't see it being useful unless you're ok
with re-writing all your perl components in the perl subset, which will be
substantial effort seeing one of the "do not"s are "regex" and we simply
don't have the expressive feature set of string functions other languages
have because we defer them to regex ...

Also, "String interpolation" is also on the list of "Do nots", which is
rather limiting also. There's quite a lot of things on the "Do not" list
that people who do a lot of perl work will be relying on.

RPerl is probably going to be fine if you're wanting to write some very low
level stuff without using dependencies that does a lot of simple
mathematical things which you need to do very quickly, the sorts of
usecases people might otherwise reach for PDL to get past perls native
speed impediments.

But based on my understanding its just not "There" yet for general purpose,
( and it may never be "there" to all/any of some peoples expectations )

-- 
Kent

*KENTNL* - https://metacpan.org/author/KENTNL
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.pm.org/pipermail/wellington-pm/attachments/20150806/953fe1bd/attachment.html>


More information about the Wellington-pm mailing list