An FSA that behaves strange

[GoogleCodeExporter]

What steps will reproduce the problem?
1st case:
1. load stack strange.fst
2. view net  (*)
3. write prolog strange.prolog
4. read prolog strange.prolog
5. view net  (**)

2nd case:
1. Compile hun_cg2.rlx with fomacg
2. load stack strange.fst
3. define Sentence
4. read prolog R_1_43.prolog
5. define Rule
6. read regex Sentence .o. Rule;  (*)
7. read regex Sentence;
8. random-upper  (**)
9. pop
10. read regex Rule;
11. apply down (**)


What is the expected output? What do you see instead?
1st case:
The net in (*) is linear; the net in (**) is star-shaped.

2nd case:
The composition in step 6 (*) does nothing -- Sentence remains unchanged. 
However, if you apply down Rule on Sentence's random-upper (one word, since 
it's linear), Rule correctly #X#'s out the non-pronoun readings of "<az>".

What version of the product are you using? On what operating system?
0.9.17alpha
Latest fomacg.

Please provide any additional information below.

Original issue reported on code.google.com by [email protected] on 30 Jul 2013 at 2:45

Attachments:

• strange.fst

[GoogleCodeExporter]

The fomacg part is so that you can reproduce the problem. The point is: there 
is sentence where apply down and .o. produces different results, even though 
they shouldn't.

Original comment by [email protected] on 30 Jul 2013 at 2:46

[GoogleCodeExporter]

OK, disregard the fomacg part, it was due to the fact that the space wasn't 
part of the wordform symbol. However, the 1st case still stands.

Original comment by [email protected] on 31 Jul 2013 at 8:41