Tomas Stupka created MINDEXER-95:
------------------------------------
Summary: suboptimal indexing execution in DefaultIndexUpdater
Key: MINDEXER-95
URL: https://jira.codehaus.org/browse/MINDEXER-95
Project: Maven Indexer
Issue Type: Bug
Affects Versions: 5.1.1
Reporter: Tomas Stupka
Attachments: mvn-indexer.diff
while profiling the maven.indexer 5.1.1 usage in NetBeans, the following turned
out:
- o.a.m.index.updater.IndexDataReader
too slow reading of the gzipped file - can be significantly improved via
DataInputStream initialization
- o.a.m.index.context.NexusAnalyzer
too much time spent in LetterOrDigitTokenizer.<init> - can be improved by
reusing the CharTokenizer-s
- not neccessary optimize calls
.optimize() seems to have a minor effect on query times, but is
disproportionally expensive during index creation time. In case you do not want
to remove it, please at least consider an option for the client code to disable
optimize calls.
- o.a.m.index.context.NexusIndexWriter
slight improvements in indexing time where achieved by tweaking the
NexusIndexWriter config. In case you do not want to change them, please
consider a way for the client code to have more control about the writer
configuration.
for particular cases and fix suggestion in code see also attached diff file ...
--
This message was sent by Atlassian JIRA
(v6.1.6#6162)
