Multiple testruns for the same machine, buildid and testdate currently produce strange results in the graph display (see URL for screenshot). This should be corrected by either adding support to store multiple data sets for same buildid, machine and rundate (preferred) or overwriting existing data.

Working on a basic fix which, if data is found for a run with the same time stamp/machine name key, the old set of data is removed allowing the new data to be applied to the db cleanly.

Should mention that the patch only works if the assumption that discrete data sets start at point 0 (ie, (0, data), (1, nextdata), (2, nextnextdata), etc).

Comment on attachment 322027 [details] [diff] [review] put a stop to dupes Looks ok to me, but I'd need a bit of time to set up a server and test it...

Comment on attachment 322027 [details] [diff] [review] put a stop to dupes Alice - if this is holding you up, I say go for it -- patch looks good to me.

Checking in collect.cgi; /cvsroot/mozilla/webtools/new-graph/collect.cgi,v <-- collect.cgi new revision: 1.9; previous revision: 1.8 done Checking in bulk.cgi; /cvsroot/mozilla/webtools/new-graph/bulk.cgi,v <-- bulk.cgi new revision: 1.4; previous revision: 1.3 done Will still require merge/check in with hg graph repository.

Pushed to graphs-stage.m.o. Will let it bake overnight before push to graphs.m.o.

Tests on stage looked fine. Let's get this pushed to production.

[root@dm-graphs01 graphs]# cvs update bulk.cgi collect.cgi P bulk.cgi P collect.cgi Updated. These are immediately live, too. (Not sure if I should close this ticket or pass it back? I'm passing it back just in case there is more action required.)

We'll have to keep this open till we get it checked in on the hg side as well.

Same as before, but for the hg repository of the graph code base.

Comment on attachment 326777 [details] [diff] [review] [Checked in] put a stop to dupes (against hg 86:b6c376afc4cb) Checked in to hg revision 278e7148ac65.