Test runs terminating if a codec throws exception?

52 views
Skip to first unread message

Tatu Saloranta

unread,
Mar 20, 2014, 1:51:37 PM3/20/14
to java-serializat...@googlegroups.com
Looks like problems with codecs are not caught any more -- it appears like the whole test run just gets quietly terminated, if a codec throws an exception for (de)serialization. This is probably not intentional, but I was curious if this is something others have seen? I am trying to add one more serializer, and debugging to see the problem has been bit slow going. I suspect there are other hidden failure modes, wrt arguments passed (iterations, runtime) -- maybe all exception output is just caught and swallowed somewhere?

-+ Tatu +-

Rüdiger Möller

unread,
Mar 21, 2014, 3:13:52 AM3/21/14
to java-serializat...@googlegroups.com
I the original branch it "works" like this:
When an exception is thrown now result is reported. The script accumulating the results then detects this and the serializer is excluded from the results. However there is output at system.err reporting this. However maybe there is a bug in this handling (e.g. exception caught and results reporte despite having an error)

from run-bench.sh:

# find files with no numbers => errors
echo ""
echo "====================================================================================="
echo "errors:"
find ./results/tmp/. -print -type f -name "*.txt" -exec tail -1 {} \; | grep -B 1 create
echo "====================================================================================="
echo ""

Rüdiger Möller

unread,
Mar 21, 2014, 3:21:39 AM3/21/14
to java-serializat...@googlegroups.com
note also you can run a single serializer isolated by passing its name to the run script. Previous results are kept from previous full runs


Am Donnerstag, 20. März 2014 18:51:37 UTC+1 schrieb cowtowncoder:

Kannan Goundan

unread,
Mar 21, 2014, 3:27:27 AM3/21/14
to java-serialization-benchmarking
Because of some folder name reshuffling, I had broken the error reporting.  But it should have been fixed in my commit on Monday.


--
You received this message because you are subscribed to the Google Groups "java-serialization-benchmarking" group.
To unsubscribe from this group and stop receiving emails from it, send an email to java-serialization-be...@googlegroups.com.
To post to this group, send email to java-serializat...@googlegroups.com.
Visit this group at http://groups.google.com/group/java-serialization-benchmarking.
For more options, visit https://groups.google.com/d/optout.

Tatu Saloranta

unread,
Mar 21, 2014, 11:30:20 AM3/21/14
to java-serializat...@googlegroups.com
Thanks -- will try to to isolate my problem further, whatever it is. So far no luck but this should help.

-+ Tatu +-


Tatu Saloranta

unread,
Mar 22, 2014, 1:57:09 AM3/22/14
to java-serializat...@googlegroups.com
Finally realized that the failure message does get printed to the results file, and not to console. So I can resolve the specific problem. The problem of rest of the run failing may have been resolved as per Kannan's comment.

-+ Tatu +-
Reply all
Reply to author
Forward
0 new messages