parse count is off-by-1 in a corner case. about link-grammar HOT 6 CLOSED

I fixed the mem over-run in commit 405aaf9

from link-grammar.

I think the mem over-run was exposed by the hebrew dict, since it does not use a left/right wall

from link-grammar.

Indeed. But it still doesn't explain why this is considered by the code to be a valid linkage.

from link-grammar.

Well, the miscount is happening at line 210 of count.c but I don't see a fix. The test there is that: if two words are next to each other, and there are no dangling connectors, then everything must be good and so it counts 1. But in this case, there are no connectors at all ... which is an equivalent condition.

from link-grammar.

also: the bug will go away if you add left, right walls to he, because that way, there will always be at least one link, from left to the right wall....

from link-grammar.

$ link-parser he
linkparser> x
Found 1 linkage (1 had no P.P. violations)
Unique linkage, cost vector = (UNUSED=1 DIS= 0.00 LEN=0)

[x]

This problem (of indicating a complete linkage when there is none) doesn't exists by now.
But another one exists instead (a duplicate linkage):

linkparser> x
No complete linkages found.
Found 2 linkages (2 had no P.P. violations) at null count 1
	Linkage 1, cost vector = (UNUSED=2 DIS= 0.00 LEN=0)

LEFT-WALL [x]

Press RETURN for the next linkage.
linkparser> 
	Linkage 2, cost vector = (UNUSED=2 DIS= 0.00 LEN=0)

LEFT-WALL [x]

This problem (of duplicate linkages resulting from "empty disjuncts") has been solved in PR #1042 (waiting for application).

from link-grammar.