Re: Deprecation of flush in IndexWriter

2008-04-13 Thread Michael McCandless
Shay Banon <[EMAIL PROTECTED]> wrote: > That what was I was hoping to get to eventually. I agree that flush can do > much more (and be better renamed to prepare). If we can get something like > that, it would be great. I'll work through a patch... > Just a note regarding two phase transactio

Re: Deprecation of flush in IndexWriter

2008-04-12 Thread Shay Banon
t; concurrent merges completes it would also be blocked from committing > the changes to the index until commit(). We should probably also > deprecate abort() and rename it to rollback(). I'll open an issue for > this... > > Mike > > Shay Banon wrote: >> >&

Re: Deprecation of flush in IndexWriter

2008-04-12 Thread Michael McCandless
t the progress you guys make is amazing!. I would still like to ask a quick question regarding deprecation of flush in IndexWriter. I think that there are cases where flush is needed. For example, in trying to create a two phase (or as close as possible to one) commit. The flush can be used

Deprecation of flush in IndexWriter

2008-04-11 Thread Shay Banon
Hi, I was just looking a bit at the trunk. First, let me say that the progress you guys make is amazing!. I would still like to ask a quick question regarding deprecation of flush in IndexWriter. I think that there are cases where flush is needed. For example, in trying to create a two phase