the motivation for this commit is long and detailed and has been
documented externally here:
https://gist.github.com/mschoch/5cc5c9cf4669a5fe8512cb7770d3c1a2
the core of the changes are:
1. recognize that collector/searcher need only a fixed number
of DocumentMatch instances, and this number can be determined
from the structure of the query, not the size of the data
2. knowing this, instances can be allocated in bulk, up front
and they can be reused without locking (since all search
operations take place in a single goroutine
3. combined with previous commits which enabled reuse of
the IndexInternalID []byte, this allows for no allocation/copy
of these bytes as well (by using DocumentMatch Reset() method
when returning entries to the pool
instead of separate DocumentMatch/DocumentMatchInternal
rules are simple, everything operates on the IndexInternalID field
until the results are returned, then ID is set correctly
the IndexInternalID field is not exported to JSON
IndexInternalID is now []byte
this is still opaque, and should still work for any future
index implementations as it is a least common denominator
choice, all implementations must internally represent the
id as []byte at some point for storage to disk
index id's are now opaque (until finally returned to top-level user)
- the TermFieldDoc's returned by TermFieldReader no longer contain doc id
- instead they return an opaque IndexInternalID
- items returned are still in the "natural index order"
- but that is no longer guaranteed to be "doc id order"
- correct behavior requires that they all follow the same order
- but not any particular order
- new API FinalizeDocID which converts index internal ID's to public string ID
- APIs used internally which previously took doc id now take IndexInternalID
- that is DocumentFieldTerms() and DocumentFieldTermsForFields()
- however, APIs that are used externally do not reflect this change
- that is Document()
- DocumentIDReader follows the same changes, but this is less obvious
- behavior clarified, used to iterate doc ids, BUT NOT in doc id order
- method STILL available to iterate doc ids in range
- but again, you won't get them in any meaningful order
- new method to iterate actual doc ids from list of possible ids
- this was introduced to make the DocIDSearcher continue working
searchers now work with the new opaque index internal doc ids
- they return new DocumentMatchInternal (which does not have string ID)
scorerers also work with these opaque index internal doc ids
- they return DocumentMatchInternal (which does not have string ID)
collectors now also perform a final step of converting the final result
- they STILL return traditional DocumentMatch (with string ID)
- but they now also require an IndexReader (so that they can do the conversion)
This optimization changes the search.Search.Next() interface API,
adding an optional, pre-allocated *DocumentMatch parameter.
When it's non-nil, the TermSearcher and TermQueryScorer will use that
pre-allocated *DocumentMatch, instead of allocating a brand new
DocumentMatch instance.
this lays the foundation for supporting the new firestorm
indexing scheme. i'm merging these changes ahead of
the rest of the firestorm branch so i can continue
to make changes to the analysis pipeline in parallel
refactor to share code in emulated batch
refactor to share code in emulated merge
refactor index kvstore benchmarks to share more code
refactor index kvstore benchmarks to be more repeatable
more things can return error now
in a couple of places we had to swallow errors because they didn't
fit the existing API. in these case and proactively in a few
others we now return error as well.
also the batch API has been updated to allow performing
set/delete internal within the batch
1. text analysis is now done before the write lock is acquired
2. there is now a pool of analysis workers
3. the size of this pool is configurable
4. this allows for documents in a batch to be analyzed concurrently
as a part of benchmarking these changes i've also introduce a new
null storage implementation. this should never be used, as it
does not actualy build an index. it does however let us go
through all the normal indexing machinery, without incuring
any indexing I/O. this is very helpful in measuring improvements
made to the text analsysis pipeline, which are often overshadowed
by indexing times in benchmarks actually building an index.
In the index/store package
introduce KVReader
creates snapshot
all read operations consistent from this snapshot
must close to release
introduce KVWriter
only one writer active
access to all operations
allows for consisten read-modify-write
must close to release
introduce AssociativeMerge operation on batch
allows efficient read-modify-write
for associative operations
used to consolidate updates to the term summary rows
saves 1 set and 1 get op per shared instance of term in field
In the index package
introduced an IndexReader
exposes a consisten snapshot of the index for searching
At top level
All searches now operate on a consisten snapshot of the index
this started initially to relocate highlighting into
a self contained package, which would then also use
the registry
however, it turned into a much larger refactor in
order to avoid cyclic imports
now facets, searchers, scorers and collectors
are also broken out into subpackages of search