This is Issue 949 moved from a Google Code project.
Added by 2012-07-14T01:35:47.000Z by e...@hrz.tu-chemnitz.de.
Please review that bug for more context and additional comments, but update this bug.
Original labels: Type-Defect, Priority-Medium, v1.3.0
<b>What steps will reproduce the problem?</b>
1. create a new database using DB_CREATE
2. open the database using DB_OPEN
3. execute COMMAND with "create class foo" (will also create cluster "foo")
5. execute COMMAND with "alter class foo name bar" (also renames cluster "foo" to "bar")
7. execute COMMAND with "drop class bar" (will also delete cluster "bar")
8. execute DB_RELOAD or DB_OPEN
<b>What is the expected output? What do you see instead?</b>
After the class has been deleted DB_OPEN & DB_RELOAD should only send the existing clusters.
Instead both commands send 7 as the number of clusters, 6 clusters can be read ("internal", "index", etc.),
the 7th cluster cannot be read since there are no more bytes left (I guess this was cluster "bar").
Any chance of a 1.2.x release or similar with this fixed? I'd hate to use 1.3-SNAPSHOT for production and it seems foolish to be hand-patching 1.2.0 each time. With this bug the 1.2 binary protocol is totally useless -- it's impossible to work around this bug because one can never be sure how many clusters to read off the socket... I can confirm this is, infact, fixed in 1.3+.
we release patches for customers only. In your case if you don't want to use the 1.3.0-SNAPSHOT why don't get the sources tagged 1.2.0 and apply the fix by your self?
where can I find the patch?
It's suppose to be in 1.3