I committed LUCENE-6970 and backported LUCENE-2229. I agree it would be bad to have changes in Lucene 5.3.1 that are not in 5.4.1, should we just cancel this 5.3.1 release and merge the bug fixes that we are interested in in the 5.4 branch? I would like to build a RC soon (I was initially thinking of tomorrow afternoon EU time) so that we can ship the corruption fix from LUCENE-6970 to our users soon as well.
Le lun. 11 janv. 2016 à 18:19, Adrien Grand <[email protected]> a écrit : > Thanks David for the explanation, I'll backport it as well then. > > Le lun. 11 janv. 2016 à 17:59, [email protected] < > [email protected]> a écrit : > >> Adrien, Mike: Thanks for working on a 5.4.1 release. >> >> IMO the bug you pointed out (LUCENE-2229) should be back-ported because I >> can't imagine someone depending on this behavior. Its ultimately a bug in >> the sizing of the fragment, and furthermore only occurs in infrequent >> circumstances. >> >> RE Solr: The 5.3.1 changes ought to move to 5.4.1, since these changes >> post-dated 5.4.0. Anshum what's the deal with 5.3.1? >> >> ~ David >> >> On Mon, Jan 11, 2016 at 10:48 AM Adrien Grand <[email protected]> wrote: >> >>> All changes in the "Bug Fixes" section of 5.5 of Lucene have been >>> backported to 5.4.1 but >>> https://issues.apache.org/jira/browse/LUCENE-2229 (thanks Mike for the >>> help). The reason is that it changes the behaviour of the highlighter and >>> I'm afraid that some users might be relying on or expecting this bug, so >>> I'd rather fold it into a minor release than a bugfix release. >>> >>> I'm not really qualified to know if there are things worth backporting >>> on the Solr side, so if you feel like some issues should be backported for >>> this bugfix release, feel free to go ahead. >>> >> -- >> Lucene/Solr Search Committer, Consultant, Developer, Author, Speaker >> LinkedIn: http://linkedin.com/in/davidwsmiley | Book: >> http://www.solrenterprisesearchserver.com >> >
