Rico Landefeld (JIRA) wrote: > [ > https://issues.apache.org/jira/browse/UIMA-1530?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel > ] > > Rico Landefeld updated UIMA-1530: > --------------------------------- > > Attachment: UIMA-1530.patch > > This patch adds a random to the index name to make it unique in multithreaded > scenarios. > > This impl uses a hash set which stores each used random (0- 1 million). I'm not sure what the "random" part of this is providing. Would using a simple incrementing integer also work?
-Marshall >> Index naming is not unique in multithreaded scenarios >> ----------------------------------------------------- >> >> Key: UIMA-1530 >> URL: https://issues.apache.org/jira/browse/UIMA-1530 >> Project: UIMA >> Issue Type: Bug >> Affects Versions: 2.3S >> Reporter: Rico Landefeld >> Fix For: 2.3S >> >> Attachments: UIMA-1530.patch >> >> >> At the moment lucas appends the host name and the pid to the index name to >> create unique indexes for each IndexWriter instance. Thats not enough, >> because there can be more then one index writer in multithreaded scenarios. >> > >
