[developers] PET 0.99.12 Core Dumps in Packing Mode

Eric Nichols eric-n at is.naist.jp
Tue Apr 25 13:21:28 CEST 2006


Greetings,

Could you provide a little more information on your config and the
version of ECL used?
Thank you for your assistance.

Eric

On 4/25/06, Berthold Crysmann <crysmann at dfki.de> wrote:
> Eric Nichols wrote:
>
> >Greetings,
> >
> >I have been playing around with the new release of PET (thanks bernd
> >and stephen :-)
> >and I think I have found a bug. It seems that cheap will core dump
> >when loading a grammar with the --packing flag enabled. I have
> >verified that this occurs in both ERG (2005/10/30) and JACY
> >(2005/11/30). This appears to have something to do with MRS because it
> >only occurs if I enable MRS mode when PET is compiled. However cheap
> >fails to read the grammar successfully REGARDLESS OF WHETHER THE --MRS
> >FLAG IS SET.
> >
> >
> >
> I can disconfirm that hypothesis. German grammar runs with packing
> enabled and MRS output.
> Platform is 32bit linux running FC5, compiled  by Bernd under 32bit FC4.
>
> B
>
> >Eric
> >
> >eric-n at thinkpad:~/delphin/grammars/jp051130 [18:58:13] [22]
> >:( cheap -packing ~/delphin/grammars/en051005/english
> >reading `/home/eric-n/delphin/grammars/en051005/pet/english.set'...
> >including `/home/eric-n/delphin/grammars/en051005/pet/common.set'...
> >`/home/eric-n/delphin/grammars/en051005/pet/global.set'...
> >loading `/home/eric-n/delphin/grammars/en051005/english.grm' (LinGO
> >(05-Sept-05)) reading ME model
> >`/home/eric-n/delphin/grammars/en051005/rondane.mem'... [11846
> >features] zsh: segmentation fault  cheap -packing
> >~/delphin/grammars/en051005/english
> >eric-n at thinkpad:~/delphin/grammars/jp051130 [18:58:23] [23]
> >:( cheap -packing ~/delphin/grammars/jp051120/japanese
> >reading `/home/eric-n/delphin/grammars/jp051120/pet/japanese.set'...
> >loading `/home/eric-n/delphin/grammars/jp051120/japanese.grm' (Jacy
> >(2005-11-20)) reading ME model
> >`/home/eric-n/delphin/grammars/jp051120/LXD-DEF-6.jp051120.mem'...
> >[60508 features] Ignoring entry with unknown rule `$-*-' in irregular
> >forms
> >Ignoring entry with unknown rule `$this' in irregular forms
> >Ignoring entry with unknown rule `$the' in irregular forms
> >Ignoring entry with unknown rule `$currently' in irregular forms
> >Ignoring entry with unknown rule `$c2stem' in irregular forms
> >Ignoring entry with unknown rule `$---' in irregular forms
> >zsh: segmentation fault  cheap -packing ~/delphin/grammars/jp051120/japanese
> >eric-n at thinkpad:~/delphin/grammars/jp051130 [18:58:46] [24]
> >:(
> >eric-n at thinkpad:~/delphin/grammars/jp051130 [18:59:36] [25]
> >:( cheap
> >parse_options(): expecting name of grammar to load
> >cheap version 0.99.12
> >usage: `cheap [options] grammar-file'; valid options are:
> >  `-tsdb[=n]' --- enable [incr tsdb()] slave mode (protocol version = n)
> >  `-nsolutions[=n]' --- find best n only, 1 if n is not given
> >  `-verbose[=n]' --- set verbosity level to n
> >  `-limit=n' --- maximum number of passive edges
> >  `-memlimit=n' --- maximum amount of fs memory (in MB)
> >  `-no-shrink-mem' --- don't shrink process size after huge items
> >  `-no-filter' --- disable rule filter
> >  `-qc-unif=n' --- use only top n quickcheck paths (unification)
> >  `-qc-subs=n' --- use only top n quickcheck paths (subsumption)
> >  `-compute-qc[=file]' --- compute quickcheck paths (output to file,
> >                           default /tmp/qc.tdl)
> >  `-compute-qc-unif[=file]' --- compute quickcheck paths only for
> >unificaton (output to file,
> >                           default /tmp/qc.tdl)
> >  `-compute-qc-subs[=file]' --- compute quickcheck paths only for
> >subsumption (output to file,
> >                           default /tmp/qc.tdl)
> >  `-mrs[=mode]' --- compute MRS semantics (in specified mode)
> >  `-key=n' --- select key mode (0=key-driven, 1=l-r, 2=r-l, 3=head-driven)
> >  `-no-hyper' --- disable hyper-active parsing
> >  `-no-derivation' --- disable output of derivations
> >  `-rulestats' --- enable tsdb output of rule statistics
> >  `-no-chart-man' --- disable chart manipulation
> >  `-default-les' --- enable use of default lexical entries
> >  `-lattice' --- word lattice parsing
> >  `-server[=n]' --- go into server mode, bind to port `n' (default: 4711)
> >  `-k2y[=n]' --- output K2Y, filter at `n' % of raw atoms (default: 50)
> >  `-k2y-segregation' --- pre-nominal modifiers in analogy to reduced relatives
> >  `-one-meaning[=n]' --- non exhaustive search for first [nth]
> >                         valid semantic formula
> >  `-yy' --- YY input mode (highly experimental)
> >  `-failure-print' --- print failure paths
> >  `-no-fullform-morph' --- disable full form morphology
> >  `-pg' --- print grammar in ASCII form
> >  `-nbest' --- n-best parsing mode
> >  `-packing[=n]' --- set packing to n (bit coded; default: 7)
> >  `-log=[+]file' --- log server mode activity to `file' (`+' appends)
> >  `-tsdbdump directory' --- write incr[tsdb] item, result and parse
> >files to `directory'
> >  `-jxchgdump directory' --- write jxchg/approximation chart files to
> >`directory'
> >  `-partial' --- print partial results in case of parse failure
> >  `-results=n' --- print at most n (full) results
> >  `-tok=(string|yy|yy_counts|xml|xml_counts)' --- select input method
> >(default `string')
> >  `-comment-passthrough[=1]' --- allow input comments (-1 to suppress output)
> >eric-n at thinkpad:~/delphin/grammars/jp051130 [18:59:38] [26]
> >:(
> >
> >--
> >--Eric Nichols
> >
> >
> >
>
>


--
--Eric Nichols




More information about the developers mailing list