We patched in the override of toArray and are still seeing the same sorts of issues.
Thanks for the help, Ben On Thu, Jun 10, 2010 at 3:05 PM, Jim Idle <[email protected]> wrote: > http://markmail.org/message/72lkr7lxgozw2axa > > > > J > > > > Jim > > > > *From:* [email protected] [mailto:[email protected]] *On > Behalf Of *Ben McCann > *Sent:* Thursday, June 10, 2010 3:03 PM > > *To:* Terence Parr > *Cc:* [email protected] > *Subject:* Re: [antlr-dev] Non-deterministic output > > > > I downloaded the source > distribution<http://antlr.org/download/antlr-3.2.tar.gz>and made that edit, > but Maven bombed out on me saying "The scm url cannot be > null." > > Thanks, > Ben > > On Thu, Jun 10, 2010 at 2:17 PM, Terence Parr <[email protected]> wrote: > > Hi Ben, > > Do me a favor and add this method > > @Override > public Object[] toArray() { > return elements.toArray(); > } > > to OrdereredHashSet in org.antlr.misc. > > I think that is a source of some nondeterminism. it helps in v4. Let me > know if this helps for you. > > Ter > > On Jun 9, 2010, at 10:34 AM, Ben McCann wrote: > > > I'm not sure what you're referring to when you say timestamps. The issue > is that the code comments keep changing. Would it be possible to just use a > List instead of a Set? That should solve the issue. > > > > Thanks, > > Ben > > > > > > $ diff {/tmp/tmpaX6PX1,/tmp/tmp2-04O_}/JsonTemplateParser.java > > 307c307 > > < // elements: value, DOC_COMMENT, startCommand > > --- > > > // elements: startCommand, value, DOC_COMMENT > > 521c521 > > < // elements: DOC_COMMENT, entry, startCommand > > --- > > > // elements: startCommand, entry, DOC_COMMENT > > 1031c1031 > > < // elements: DOC_COMMENT, entry > > --- > > > // elements: entry, DOC_COMMENT > > 1161c1161 > > < // elements: DOC_COMMENT, key, value > > --- > > > // elements: key, value, DOC_COMMENT > > 1365c1365 > > < // elements: value, DOC_COMMENT > > --- > > > // elements: DOC_COMMENT, value > > > > > > > > > > On Wed, Jun 9, 2010 at 10:21 AM, Terence Parr <[email protected]> > wrote: > > yeah, there are Sets in there. i wonder why (how?) java alters the order > for same computation across runs. Come to think of it, i'm chasing > something similar in ANTLR v4 unit tests. I thought it was static data > somewhere but could be some weird Set thing. > > > > damn java. grr.... not sure how easily i can help you here... > > > > Ter > > > > On Jun 9, 2010, at 10:15 AM, Ben McCann wrote: > > > >> We're using 3.2. The only thing I've noticed changing between runs thus > far is the code comments. I should also mention the input is not changing > at all. I'm guessing it's something along the lines of using an Iterator > from a Set instead of List so that there's no ordering guarantee. > >> > >> Thanks, > >> Ben > >> > >> > >> On Wed, Jun 9, 2010 at 10:03 AM, Terence Parr <[email protected]> > wrote: > >> wow. that's weird. surely the grammar changes between runs? Computers > are usually good about determinism ;) > >> What version of antlr? > >> Ter > >> On Jun 8, 2010, at 9:52 PM, Ben McCann wrote: > >> > >>> Hi, > >>> It appears that ANTLR is generating varying output between runs. This > causes problems for us when trying to diff to see if anything has changed. > Could someone file a bug for this? I was not able to register for the > JIRA. > >>> > >>> Thanks, > >>> Ben McCann > >>> Software Engineer > >>> Google Inc. > >>> > >>> > >>> $diff ~/PathParser.java.1 ~/PathParser.java.2 > >>> 560c560 > >>> < // elements: Variable, expr > >>> --- > >>> > // elements: expr, Variable > >>> > >>> _______________________________________________ > >>> antlr-dev mailing list > >>> [email protected] > >>> http://www.antlr.org/mailman/listinfo/antlr-dev > >> > >> > > > > > > > > _______________________________________________ > antlr-dev mailing list > [email protected] > http://www.antlr.org/mailman/listinfo/antlr-dev > >
_______________________________________________ antlr-dev mailing list [email protected] http://www.antlr.org/mailman/listinfo/antlr-dev
