Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2014-10-04

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

All times shown according to UTC.

Time Nick Message
03:21 aredridel joined #marpa
03:43 CQ_ joined #marpa
07:08 jdurand joined #marpa
07:09 shadowpaste "jdurand" at 88.160.190.154 pasted "leak ?" (38 lines) at http://scsys.co.uk:8002/427782
07:10 jdurand Jeffrey; c.f. paste upper: suppose I do a marpa_r_earleme_complete that failed, and want to abort immediately the parse, i;e. it is followed by a marpa_r_unref(). I observe a memory leak. Am I doing it wrong ?
07:12 jdurand This was with Marpa-R2-2.088000 - thx
08:21 jeffreykegler joined #marpa
08:22 jeffreykegler jdurand: re http://irclog.perlgeek.de/marpa/2014-10-04#i_9454635 -- could you file this as a Github issue, against Marpa::R2
08:54 lwa joined #marpa
10:25 jeffreykegler joined #marpa
12:56 jdurand joined #marpa
12:57 jeffreykegler joined #marpa
12:59 jdurand Jefrey, what is the native order of items when fetching progress report using marpa_r_progress_item () ?
13:43 jeffreykegler jdurand: The order is subject to change, so code should not rely on it.
14:17 flaviu1 joined #marpa
14:22 jdurand jeffreykegler: ok, thx
15:42 jeffreykegler jdurand: I've attempted a fix in commit 606b4d4ecafdcbb5f2c66d1b07e0ab68b2130931
15:43 jeffreykegler I have tested it for regression, but *not* to see if it actually fixes your leak.  Could you check this?
16:02 jdurand make releng failed with "MANIFEST appears to be out of sync with the distribution"
16:02 jdurand what is the make target to do the tarball
16:17 jeffreykegler Usually it will say why it is out of sync -- files missing, etc.
16:19 jeffreykegler Did it say what the problem is?
16:20 jdurand Not in MANIFEST: new.pl
16:20 jdurand Not in MANIFEST: prepare_build_dir.pl
16:20 jdurand That's all
16:21 jeffreykegler Delete those files and rerun
16:26 jdurand Yep - worked - they reappeared at releng though via a git status - but the tarball is here - checking now the leak fix
16:28 jdurand Ah... you changed "core" to "engine" -;
16:28 jeffreykegler "core" is a bad choice of name for a file on a UNIX system
16:29 jeffreykegler You're not installing in a fresh directory?  All sorts of stuff has changed since 088
16:31 jeffreykegler Or I guess your test needs to know the directory names -- sorry
16:33 jdurand No pb I am adapting my perl script at this moment
16:37 jdurand Yes, leak has disappeared - many thx
16:41 jeffreykegler I just closed the Github issue
16:41 jdurand Well done - thx again
16:45 jdurand I wonder. to detect that a parse is failing, is that correct to assume that the fact that marpa_r_earleme_complete() failed is enough & is MARPA_ERR_PARSE_EXHAUSTED always an indication of parsing failure?
16:47 jeffreykegler marpa_r_earleme_complete() will fail if called at a point where you do not have a grammatically acceptable prefix.
16:48 jeffreykegler Re exhaustion and success and failure: https://github.com/jeffreykegler/Marpa--R2/blob/master/cpan/pod/Exhaustion.pod
16:49 jeffreykegler I just wrote it for the last release ...
16:49 jeffreykegler I hope it "exhausts" the topic. :-)
16:51 jdurand Done reading it. I am not sure about your statement for XML/HTML. Grammar itself does not enforce balanced tags. This is a semantic check.
16:52 jdurand ./.. which usually is done at lexing phase. But not necessarly.
16:53 jeffreykegler My way of doing XML/HTML tags was to do a 1st pass to census the tags --
16:53 jeffreykegler Then I wrote a grammar and used that for the next phase, the actual parse.
16:54 jeffreykegler "wrote" meaning "generated programmatically"
16:54 jeffreykegler Another technique you just cannot use with LALR or recursive descent
16:54 jdurand Ah, yes. I remember. This was another approach that justify your statements. In fact it depends on what grammar is used behind, of course. The gem of W3C grammar does not have this enforcement, this is what I meant.
16:57 jeffreykegler I think what I say in the Exhaustion doc is correct, if only because it's vague enough. :-)
16:58 jdurand Sure -;
17:06 jdurand AFK
17:11 jeffreykegler By the way, those who want to compare my approach to and writeup of exhaustion to the Grune&Jacobs text, their writeup is on pp. 93-94 (2nd ed.).
18:44 flaviu1 joined #marpa
22:38 ronsavage joined #marpa

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