June 29, 2011

#perl6 2011-06-29

Meta


(TimToady)
moritz: new idea, after talking about the no-longer-done mailing list summaries, we thought it would be cool if someone, while reading the ir clogs, could just flag important things as a pseudo-summary

(moritz)
TimToady: re http://irclog.perlgeek.de/perl[…]1-06-29#i_4021067 it can certainly be done, though I'm not yet sure if it should include authentication, or just a global (not perl nick) flag per line


News


(colomon)
moritz: http://www.harmonyware.com/perl/p6numerics/ (second draft)

(PerlJam)
colomon: that point about the Bridge not specifying the type was a revelation to me.  I hadn't thought much about it before.

(masak)
TimToady: I added a paragraph about <...> to http://strangelyconsistent.org[…]d-data-structures based on your feedback.

(masak)
blog post! http://strangelyconsistent.org[…]22-format-strings

(colomon)
TimToady: I've updated the talk with the changes to the mandel sub you suggested and additional comments :)



Language


(TimToady)
getc is a stdio function

(TimToady)
it doesn't control what goes into the buffer


(TimToady)
the compounds of s,m,q, etc are just that letter plus the switch

(TimToady)
so ss turns into s:s and sss would turn into s:ss

(TimToady)
otoh, if we slurp up a bunch of keywords starting with s, people will hat us

(TimToady)
and we proably shouldn't make s:foo -> sfoo a productive mechanism anyway

(TimToady)
there is a hypothetical 'quote' declarator

(TimToady)
so we could just arbitrary declare some common ones, like qqx


(jlaire)
S05 has a few instances of C<s//>, I changed them all to C<s///>: http://laire.fi/t/S05.patch

(dalek)
specs: s// -> s/// from jlaire++


([Coke])
ROODE was pondering how nice it would be if he had $foo ?= 'a' : 'b' in 5.

(sorear)
$foo ?? 'a' !!= 'b' would work, but is specifically forbidden


(pmichaud)
I don't agree with jnthn/masak's proposal (or some of the details of the resulting specpatch)

(pmichaud)
I especially don't like the 1=>  stuff.

(sorear)
the only change I would have made personally is to make :x opt-in

(TimToady)
well, would it help to say that the :x form uses the names in the proto, but the actual multis can use different names on the same positions?


(TimToady)
still would like to see S12:2268 happen


Internals


(pmichaud)
TimToady: what's the best model/code for me to look at to implement full ltm-semantics?

(TimToady)
well, my stuff did real LTM but not with a real dfa-ish engine, while it's my understanding that niecza does it in a theoretically purer fashion

(sorear)
AFAIR discarding fates is the main break niecza makes from Cursor

(sorear)
well not so much "discarding" as "I haven't implemented them yet, and with DFAs being <10% of parse time I see no pressing need"

(sorear)
pmichaud: the basic principle of the LTM handling (in niecza and Cursor) is that | is a || that intelligently reorders clauses

(sorear)
Cursor.cs

(sorear)
pmichaud: start from LTMPushAlts


Niecza


(sorear)
niecza now passes every substr.t test except the StrPos ones

(sorear)
btw, niecza is now passing over 3000 spectests

(dalek)
niecza: Implement term:<proceed>, Any.{isa,can,does}, we now pass given.t

(sorear)
I predict that by the time of the July release, Rakudo will be the faster one, but Niecza will have the more features

(dalek)
niecza: Implement <-> auto-rwification

(sorear)
it turned out to be very easy


Modules


(tadzik)
dukeleto, colomon: that's a bit messy. projects.list is for backwards-compatibility only, and should be buried probably. Now module-starter was added to the ecosystem long ago, but it was moritz's module. My version wasn'

(dukeleto)
tadzik: i gave module-starter some love today


Tests


(colomon)
sorear: yes, what I'm doing is changing the automatic trig test generator to get rid of the trigbase tests

(dalek)
roast: Remove the trigbase tests from the trig spectests.

(sorear)
masak: being able to test the REPL would be _so awesome_

(sorear)
mberends: the standard way to test a Perl 6 implementation is prove -e IMP test.pl


Rosetta Code


(Util)
I think Rat is only *used* in this draft task: http://rosettacode.org/wiki/Co[…]rt_decimal_number

(Util)
Rationals greatly enhance the awesomeness of http://rosettacode.org/wiki/Re[…]helon_form#Perl_6

(Util)
We get *exact* output; compare to Ruby, Groovy, and Javascript.

(TimToady)
Util: it was http://rosettacode.org/wiki/Va[…]t_sequence#Perl_6 I was thinking of

(TimToady)
shows off the Rats