lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Steven Rowe (JIRA)" <>
Subject [jira] [Commented] (LUCENE-4335) Builds should regenerate all generated sources
Date Fri, 31 Aug 2012 14:08:08 GMT


Steven Rowe commented on LUCENE-4335:

bq. I've seen JFlex reorder cases in switch statements, even when there are no .jflex source
changes, when run under different JVM versions. I recall seeing this specifically when generating
under Java5 and Java6, both Oracle JVMs on Windows.

Dawid sent me a patch to use LinkedHashMaps instead of HashMaps for the data structures emitted
as switch cases, so insertion order will be the emit order.  I committed Dawid's patch to
JFlex trunk r614.  Please update and rebuild to get the change.
> Builds should regenerate all generated sources
> ----------------------------------------------
>                 Key: LUCENE-4335
>                 URL:
>             Project: Lucene - Core
>          Issue Type: Improvement
>            Reporter: Michael McCandless
>         Attachments: LUCENE-4335.patch
> We have more and more sources that are generated programmatically (query parsers, fuzzy
levN tables from Moman, packed ints specialized decoders, etc.), and it's dangerous because
developers may directly edit the generated sources and forget to edit the meta-source.  It's
happened to me several times ... most recently just after landing the BlockPostingsFormat
> I think we should re-gen all of these in our builds and fail the build if this creates
a difference.  I know some generators (eg JavaCC) embed timestamps and so always create mods
... we can leave them out of this for starters (or maybe post-process the sources to remove
the timestamps) ...

This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see:

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message