[developers] Regular Expressions in Transfer Rules

Francis Bond bond at ieee.org
Thu Mar 2 02:00:28 CET 2017


G'day,

Thanks Woodley and Stephan for your input.

I also tried adding a new test_mark to the MRS to be transferred.  This got
> rewritten as "ja:test_mark" and did *not* end up matching the rule.  My
> hypothesis is that this is due to the final quote.  ACE’s transfer
> implementation does not (yet) fully implement the brave new world under
> which quotes and _rel suffixes are irrelevant to predicate matching, and
> when it does you will probably need to enable it with a configuration
> option.  A possible workaround would be to write "~mark\"?$".
>

This did indeed solve the problem.   another solution would be to make sure
all marks are subtypes of mark (and not use strings) then we can just match
under subsumption and avoid the regular expression altogether.


> Incidentally, how do you feel about "~mark$" matching "_mark_rel"?  To me,
> it would seem hard to justify ignoring quotes without also ignoring the
> _rel suffix.  I would be curious to know the current LOGON implementation’s
> stance on this.
>

Thanks Stephan for the explanation here.   I think moving to normalization
is worth it, even if we did end up matching "_mark_rel".   I can see
several workarounds (use a more obscure name, use types for marks, ...)
whereas the issues caused by strings and types behaving differently are
hard to predict and debug.


> If you don’t think the behavior you are observing matches the situation
> above, could you provide an example that causes the behavior so that I can
> reproduce it locally?
>

Sorry not to give an example the first time round (I sent the mail from a
different machine).

Yours,

-- 
Francis Bond <http://www3.ntu.edu.sg/home/fcbond/>
Division of Linguistics and Multilingual Studies
Nanyang Technological University
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.delph-in.net/archives/developers/attachments/20170301/59f166a6/attachment.html>


More information about the developers mailing list