https://issues.apache.org/bugzilla/show_bug.cgi?id=49835

--- Comment #5 from Andreas L. Delmelle <adelme...@apache.org> 2011-01-31 
19:23:52 EST ---
Created an attachment (id=26583)
 --> (https://issues.apache.org/bugzilla/attachment.cgi?id=26583)
Minimal Test Case - Slightly Altered

So I took a closer look at this one. No immediate fix in mind yet, but one
observation so far that could be of use in further tracking this down.

Take a look at the revised attachment. The overflow recovery/deferral mechanism
is doing exactly what it is supposed to here. The issue seems to manifest
itself in case an overflow condition is reached, and there is no 'real'
previous node yet. The auxiliary node that is added at the start, right before
the main loop in findBreakingPoints() is entered, is deactivated, and the
algorithm then tries to create a 'last forced' node to restart from later
(findBreakingPoints() line 577 -> recoverFromTooLong()). This all seems to work
fine as long as there actually is a valid node to restart from, but something
seems to go wrong in case there isn't, yet.

-- 
Configure bugmail: https://issues.apache.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.

Reply via email to