Michael McCandless

Hmm, the `.tmp` file should never be part of the file set for the actual segment? It should be transient, and then re-copied into the final result (the actual block...

And you can use the `IndexOutput.getName()` to get the `String` name of the temp file, and delete that file (prolly in a `finally` clause in case disaster strikes) using `Directory.deleteFile`...

> I am seeing CheckIndex gets a handle on an EverythingEnum (and other enums) over a test field indexed with no positions and no freqs. Hmm, does `CheckIndex` pull all...

> also -- the slowdown for AndHighHighDayTaxoFacets counters the overall trend. I wonder what's going on there. Wait -- this task got faster right? And some others got slower, e.g....

I really love this idea! And it's wild that it's net/net reducing `enwiki` index size even at higher than expected cutover to dense encoding criteria!

I think it's crazy that ES (and now OS) does not offer a streaming indexing API. The bulk indexing API is horrible -- clients must figure out 1) how many...

I hope we do not expose cluster details like "heap is low", "too many concurrent requests", etc., through this new API. Rather, the cluster should pull from the stream at...

Thanks @msokolov. This looks like a nice tool, helpful for giving demos of cool Lucene features at conferences, but it looks like consensus is we should not add it to...

It looks like there are strong objections to sharing string instances here, and there is a JVM command-line flag that may achieve similar gains for many indices X segments X...

Thanks for the reminder/ping @stefanvodita! The latest iteration looks great to me -- I'll wait for a few days to see if @zhaih has more feedback and then merge. It's...