User-Agent: Mozilla/5.0 (Windows NT 5.1; rv:2.0.1) Gecko/20100101 Firefox/4.0.1 Build Identifier: 4.0.1 I'm running two Bugzilla 4.0.1 installations on Oracle 11.2 and 10.2.0.1 databases and have found that the new duplicate bug detection never returned any potential duplicates, even when testing with a database containing a couple of bugs whose summaries are were as simple as "test one" and "test two" - entering "test" into the summary field of a new bug I consistently get no possible duplicates listed. Digging a bit deeper, I found that when I ran I dumped the SQL within possible_duplicates in Bug.pm to a temp file and ran it manually in SQL Developer, I got "ORA-00920: invalid relational operator" on the CONTAINS clause. Adjusting that to append a > 0 on the end (so the clause reads "WHERE (CONTAINS(bugs_fulltext.short_desc,'Test',1) > 0)" instead) gets past that particular error, but I then hit "ORA-00933: SQL command not properly ended", which is complaining about the "LIMIT 12" part of the query. I was able to get around this by replacing the "LIMIT 12" with a "AND ROWNUM <= 12", but that's pretty Oracle specific so most likely isn't the correct fix, but it did result in potential duplicates being listed when creating a new bug. Reproducible: Always Steps to Reproduce: 1.Enter a bug with a summary "test bug" on a Bugzilla installation running on an Oracle database (reproduced on 10.2 and 11.2 databases. 2.Start to enter a second bug and set the summary on the new bug to "test" then wait for potential duplicates to be reported. Actual Results: No duplicates will be listed underneath the summary field on the second bug. Expected Results: The first bug should be listed as a potential duplicate.
This may be a bug in our Oracle driver, the code it uses to figure out LIMIT may not be working here.
Summary: Duplicate bug detection doesn't work when using Oracle databases. → [Oracle] Duplicate bug detection doesn't work when using Oracle databases.
Bugzilla::Bug::possible_duplicates() uses LIMIT hardcoded instead of $dbh->sql_limit().
Status: UNCONFIRMED → NEW
Depends on: 22353
Ever confirmed: true
OS: Windows XP → All
Hardware: x86 → All
Target Milestone: --- → Bugzilla 4.0
Version: unspecified → 4.0.1
This fixes the problem with Oracle 10.2.
Assignee: create-and-change → LpSolit
Status: NEW → ASSIGNED
Attachment #552823 - Flags: review?(mkanat)
Comment on attachment 552823 [details] [diff] [review] patch, v1 Review of attachment 552823 [details] [diff] [review]: ----------------------------------------------------------------- ::: Bugzilla/DB/Oracle.pm @@ +177,4 @@ > my ($self, $column, $text, $label) = @_; > $text = $self->quote($text); > trick_taint($text); > + return "CONTAINS($column,$text,$label) > 0", "SCORE($label)"; Why is CONTAINS > 0 necessary? That looks like a MySQL-ism.
(In reply to Max Kanat-Alexander from comment #4) > Why is CONTAINS > 0 necessary? That looks like a MySQL-ism. Because Oracle fails without it? :)
Also, examples on the Oracle website use CONTAINS(...) > 0.
Comment on attachment 552823 [details] [diff] [review] patch, v1 Sounds good to me, then! :-)
Attachment #552823 - Flags: review?(mkanat) → review+
Committing to: bzr+ssh://email@example.com/bugzilla/trunk/ modified Bugzilla/Bug.pm modified Bugzilla/DB/Oracle.pm Committed revision 7922. Committing to: bzr+ssh://firstname.lastname@example.org/bugzilla/4.2/ modified Bugzilla/Bug.pm modified Bugzilla/DB/Oracle.pm Committed revision 7902. Committing to: bzr+ssh://email@example.com/bugzilla/4.0/ modified Bugzilla/Bug.pm modified Bugzilla/DB/Oracle.pm Committed revision 7643.
Status: ASSIGNED → RESOLVED
Last Resolved: 8 years ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.