[developers] compare-in-detail()

Emily M. Bender ebender at uw.edu
Thu Apr 25 01:58:13 CEST 2013

Dear Stephan,

In my seminar this week we were just talking about presuppositions,
including the one that comes along with the definite determiner.
So I can say with authority that if I had been using the string
"the fine system" as a definite description, that utterance would
not have suffered from presupposition failure: The referent is surely
uniquely identifiable!

But, as you well know, English orthography can be slippery, and
in fact that string of characters I typed was not the definite description
you took it for but in fact another spelling of the phrase pronounced
(I believe) "tee ess dee bee PLUS plus".

--- In all seriousness, thank you for the quick fix!


On Wed, Apr 24, 2013 at 2:10 PM, Stephan Oepen <oe at ifi.uio.no> wrote:
> dear emily,
> which of the many DELPH-IN fine systems are you using?
> it sounds like maybe [incr tsdb()], but before looking over
> recent code changes there, i would like to know for sure ...
> best, oe
> On Wed, Apr 24, 2013 at 9:50 PM, Emily M. Bender <ebender at uw.edu> wrote:
>> Dear all,
>> I have just reproduced this problem and can provide a little more
>> detail:
>> 1) It is not specific to profiles from the matrix regression test
>> set.
>> 2) It is specific to tsdb::compare-in-detail() when the intersection
>> includes `mrs'.  (Just `readings' doesn't cause problems, nor
>> `derivation', nor `tree'.)
>> 3) Loading the grammar associated with the profiles doesn't
>> make a difference.  I loaded the erg from logon/lingo/erg,
>> and then created two profiles from the `mrs' skeleton, and processed
>> each.  This should lead to identical profiles.  If the intersection
>> is set to include mrs, however, there are diffs for every single
>> item (no mrses in common).
>> 4) I tried comparing the profiles with diff, and the results files
>> do differ, but ediff-buffers() in emacs suggests that the differences
>> are in the time field.
>> 5) I used cp to create an absolutely identical copy of one of
>> the profiles, and *still* tsdb::compare-in-detail reports diffs
>> if the intersection includes mrses.
>> Since we use this functionality of the fine system in our regression
>> testing for the Grammar Matrix, the current state of affairs
>> means we can't run the regression tests at all.   For the meantime,
>> we can back up to an earlier version of LOGON.  Any clues
>> as to what might be culprit and which version we should back
>> up to?
>> Thanks,
>> Emily
>> On Sun, Apr 21, 2013 at 8:56 PM, Sanghoun Song <sanghoun at uw.edu> wrote:
>>> Deae Folks!
>>> I have a question about (tsdb::compare-in-detail ...).
>>> I have two profiles (current and gold) that consist of exactly the
>>> same files, which are used for the regression test in the Matrix
>>> systems. But if I run compare-in-detail() with the two same profiles,
>>> tsdb says they are different.
>>> For your replication, I attached two sample profiles, and a shell
>>> script (foo.sh), adapted from run_regression_test.sh in the Matrix.
>>> [$ diff current/ gold/] provides nothing, but comparison between them
>>> in tsdb shows 6 differences.
>>> FYI, I updated LOGON very recently.
>>> Sanghoun
>>> --
>>> ====================================
>>> Sanghoun Song (PhC)
>>> Dept. of Linguistics, Univ. of Washington
>>> http://corpus.mireene.com
>>> ====================================
>> --
>> Emily M. Bender
>> Associate Professor
>> Department of Linguistics
>> Check out CLMS on facebook! http://www.facebook.com/uwclma

Emily M. Bender
Associate Professor
Department of Linguistics
Check out CLMS on facebook! http://www.facebook.com/uwclma

More information about the developers mailing list