Steps to reproduce:

1. Problem first noticed upgrading from 6.0.5 to 7 Alpha 8, many active triggers were no longer active.  These were all using {TRIGGER.VALUE} to configure different trigger/recovery states instead of a recovery expression (they date from before it existed)
2. To create fresh, add an item with "icmpping" simple check only (host Flood9 in this case), have the device be down, then add a trigger as follows: 

( {TRIGGER.VALUE}=0 and last(/Flood9/icmpping)=0 ) or 
( {TRIGGER.VALUE}=1 and last(/Flood9/icmpping)=0 )
 

Result:

The result is "OK" when it should be "PROBLEM".   If you switch to a recovery expression instead, it works properly.

See screen shot attached.

Expected:

In the screen shot both should show PROBLEM, and this works in 6.0.5 (I do not have any servers between that and 7.0A8).