lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Uwe Schindler" <>
Subject RE: Landing the flex branch
Date Thu, 01 Apr 2010 12:04:19 GMT

we should think about how to merge the changes to trunk. I can try this out during the weekend,
to merge back the changes to trunk, but this can be very hard. So we have the following options:

Try a merge back: This would let flex appear as a single commit to trunk, so the history of
trunk would be preserved. If somebody wants to see the changes in the flex branch, he could
ask for them (e.g. in TortoiseSVN there is a checkbox "Include merged revisions"). If this
is not easy or fails, we can do the following:

- Create a big diff between current trunk and flex (after flex is merged up to trunk). Attach
this patch to an issue and let everybody review. After that we can apply the patch to trunk.
This would result in the same behavior for trunk, no changes lost, but all changes in flex
cannot be reviewed.
- Delete current trunk and svn move the branch to trunk (after flex is merged up to trunk):
This would make the history of flex the current history. The drawback: You losse latest trunk
changes since the split of flex. Instead you will only see the merge messages. Therefore we
should see this only as a last chance.


Uwe Schindler
H.-H.-Meier-Allee 63, D-28213 Bremen

> -----Original Message-----
> From: Michael McCandless []
> Sent: Tuesday, March 30, 2010 5:35 PM
> To:
> Subject: Landing the flex branch
> I think the time has finally come!  Pending one issue (LUCENE-2354 --
> Uwe), I think flex is ready to land.... I think the other issues with
> Fix
> Version = Flex Branch can be moved to 3.1 after we land.
> We still use the pre-flex APIs in a number of places... I think this
> is actually good (so we continue to test the back-compat emulation
> layer).  With time we can cut them over.
> After flex, there are a number of fun things to explore.  EG, we need
> to make attributes work well with codecs & indexing/searching (with
> Multi/DirReader, serailize/unserialize, etc.); we need a BytesRef +
> packed ints FieldCache StringIndex variant which should use much less
> RAM in certain cases; we should build a fast core PForDelta codec;
> more queries can cutover to operating directly on byte[] terms, etc.
> But these can all come with time...
> Thoughts/issues/objections?
> Mike
> ---------------------------------------------------------------------
> To unsubscribe, e-mail:
> For additional commands, e-mail:

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

View raw message