Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2015-11-07

| Channels | #marpa index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
01:25 Idiosyncrat *Still* working on the paper.  The longer it takes, the more I am convinced that it needs to be done.
01:26 Idiosyncrat One work I've almost eliminated from my mathematical vocabulary is "clearly".
01:27 Idiosyncrat As in "Clearly every homomorphism of a widget frobitzes some wimble"
01:28 Idiosyncrat Whenever I make a revision pass, if it's not immediately clear to me not only that whatever is "clear" is true ...
01:28 Idiosyncrat but *why* it is true, then I rewrite adding that part I could not think of instantly and off the top of my head.
01:29 Idiosyncrat I think the only survivor is "Clearly Step n derives Step n+1 in one step"
01:30 Idiosyncrat That one *did* seem clear. :-)
02:27 Idiosyncrat joined #marpa
02:39 Idiosyncrat joined #marpa
02:48 ilbot3 joined #marpa
02:48 Topic for #marpa is now Start here: http://savage.net.au/Marpa.html - Pastebin: http://scsys.co.uk:8002/marpa - Jeffrey's Marpa site: http://jeffreykegler.github.io/Marpa-web-site/ - IRC log: http://irclog.perlgeek.de/marpa/today
18:30 jdurand joined #marpa
18:33 jdurand Re http://irclog.perlgeek.de/m​arpa/2015-11-03#i_11476287 - the fact is Marpa is thread-safe is the only real requirement that as a programmer I expect from a third-party library. As said OO is just a design pattern, and nothing prevent to do OO on top of Marpa even if the perl's package is not really like this - the core engine itself must remains as it is IMHO
18:34 jdurand The only thing that I had to fight with is to redirect the Marpa logging the way I want - and the use of the say() function in Marpa is remarquabliy vicious -; I finally solved it with the famous tie() interface _user2 talked about indeed, c.f. https://github.com/jddurand/MarpaX-Role-Parameter​ized-ResourceIdentifier/blob/master/lib/MarpaX/Ro​le/Parameterized/ResourceIdentifier/MarpaTrace.pm
20:11 Idiosyncrat joined #marpa
20:12 Idiosyncrat jdurand: re http://irclog.perlgeek.de/m​arpa/2015-11-07#i_11499689
20:13 Idiosyncrat Yes, my thought was always "Is this something the user can add themselves if, and in the way that, they waht it?"
20:14 Idiosyncrat So Marpa is thread-safe, but itself makes no use of threading, for example.
20:15 Idiosyncrat (And I went to the extent of avoiding recursive subroutine calls -- hard to do in parsing and evaluation -- because some threaded environments are very tight on stack.)
20:16 Idiosyncrat (In cases where you'd ordinarily recurse, Marpa uses custom stacks that are not on the call stack.)
20:16 Idiosyncrat If I know that "say" would be trouble.  I'd certainly have avoided it.
20:17 Idiosyncrat Ironically, I started Marpa around when 5.10 went out and supported only it, on the idea that Marpa would be a long term project.
20:18 Idiosyncrat I am surprised at how long 5.8 has lasted.
20:19 Idiosyncrat Also, the 5.10 features I wanted were "say", Module::Build and smart matching.
20:19 Idiosyncrat Module::Build has since retroactively been taken out of the core.
20:20 Idiosyncrat Smart matching was retroactively declared "experimental".
20:20 Idiosyncrat And now say is problematic.  Sigh.
20:22 Idiosyncrat The only new feature I looked for in 5.10 that has not either proved more trouble than it was worth, or retroactively withdrawn is the "//" operator.
20:23 Idiosyncrat And given that vs. 5.8 support, I think now that I'd pick 5.8 support.
20:32 jdurand Hey, you did a remarquable work with Marpa with, and let's say for, the per community, and my next modules will hopefully continue to spot the light in this direction - I am working quite hard in a rewrite of a very very used module that is, quite estonishing, almost very wrong
20:32 jdurand "the perl community"
22:18 ronsavage joined #marpa
22:59 Idiosyncrat Just looked at the slides for MJD's hash talk
22:59 Idiosyncrat http://perl.plover.com/classes/H​ashHistory/samples/slide025.html
22:59 Idiosyncrat Ooops, I meant http://perl.plover.com/classes/HashHistory/
23:00 Idiosyncrat Pretty good history, but I beg to differ when he describes AVL and their teaching as "useless X"
23:01 * Idiosyncrat has replaced a word banned on this confessedly rather prissy channel with X
23:02 Idiosyncrat Libmarpa, which is substantial and makes extensive use of searches, uses only AVL's
23:02 Idiosyncrat MJD's slides do not mention that AVL's are O(log n) worst case for searches and insertions, which is better than hashes.
23:03 Idiosyncrat He's right that the worst case for hashes is rare, absent malice.
23:04 Idiosyncrat And leaves it at that making me wonder if he has somehow found a kinder, gentler, safer Internet than the one I know.
23:05 Idiosyncrat Whenever a new advisory goes out about hashes, I and Libmarpa users can just ignore it.
23:05 Idiosyncrat Plus we don't have to worry about bad luck with our hash keys.
23:07 Idiosyncrat Bad luck occurs *a lot* more than you'd hope, because in real life user generated search keys are *not* random.
23:09 Idiosyncrat MJD and his work is a major influence on my efforts, by the way.
23:09 Idiosyncrat And I envy his ability to get super-mathy in the Perl community without any blowback.

| Channels | #marpa index | Today | | Search | Google Search | Plain-Text | summary