When we repeatedly run a variant of your sample scanner
using the same input we get varying outputs. We've
traced the code to these lines in X12Document:
pf.seek(pos);
// strip the GE and GS segments and add to
outfile
while ( (ge = pf.readLine()) != null) {
Although pos (in our case) is always == 0, sometimes we
get a null pointer on first read and sometimes not.
A bug in the JDK? We're running Sun's \j2sdk1.4.2_04.
Have you encountered or fixed this problem?
BTW, I'll be out of the office tommorrow. Please
communicate with (or cc) Marianne for all
correspondence regarding this problem at:
MRives@policy-studies.com.
thx,
Owen Newnan
ONewnan@policy-studies.com
PS, we also had to mod edi4j to achieve addendum
compliance, can follow up on that another time...
Logged In: YES
user_id=1221973
I am a newbie and have no idea if this posting has been
looked at or responded to. I should however let you know
that this behavior was due to the underlying file being
sometimes empty and sometimes not. That in turn was due to
errors in IO handling in X12Document breakGroups methods --
multiple versions of a file open for output at the same
time, temp files being created with non-unique names.
This results in upredictable data corruption and/or loss
that make the product unreliable. This should be a high
priority issue for anyone using the technology.
We've found several of the bugs, but not all. We've also
applied updates for addendums and fixed other bugs. We're
in sort of a time crunch right now, but will of course
gladly share everything we've got and would appreciate
hearing anything you know about such problems.