diff options
Diffstat (limited to 'skate')
| -rw-r--r-- | skate/zipkey/batch.go | 9 | ||||
| -rw-r--r-- | skate/zipkey/zipkey.go | 2 |
2 files changed, 7 insertions, 4 deletions
diff --git a/skate/zipkey/batch.go b/skate/zipkey/batch.go index 56f6f0d..e8ba052 100644 --- a/skate/zipkey/batch.go +++ b/skate/zipkey/batch.go @@ -10,7 +10,9 @@ const ( maxGroupSize = 500 // short circuiting threshold ) -// Batcher runs reducers in parallel on batches of groups. +// Batcher runs reducers (over groups) in parallel on batches. This can have a +// significant impact on runtime, e.g. for certain datasets and operations we +// saw a reduction of 80% in processing time. type Batcher struct { Size int NumWorkers int @@ -27,7 +29,8 @@ func NewBatcher(gf groupFunc) *Batcher { return NewBatcherSize(gf, defaultBatchSize) } -// NewBatcherSize initializes a batcher with a given size. +// NewBatcherSize initializes a batcher with a given function to apply and a +// batch size. func NewBatcherSize(gf groupFunc, size int) *Batcher { batcher := Batcher{ gf: gf, @@ -60,7 +63,7 @@ func (b *Batcher) Close() error { // groupFunc before. Not thread safe. Panics if called after Close. func (b *Batcher) GroupFunc(g *Group) error { if b.closing { - panic("cannot call GroupFunc after Close") + panic("batcher: cannot call GroupFunc after Close") } b.batch = append(b.batch, g) // A few groups have an extended size, e.g. thousands of members. We short diff --git a/skate/zipkey/zipkey.go b/skate/zipkey/zipkey.go index 3e8a133..ffd33fe 100644 --- a/skate/zipkey/zipkey.go +++ b/skate/zipkey/zipkey.go @@ -32,7 +32,7 @@ type ZipRun struct { sep byte } -// New create a new ready to run ZipRun value. +// New creates a new ready to run ZipRun value. func New(r0, r1 io.Reader, kf keyFunc, gf groupFunc) *ZipRun { return &ZipRun{ r0: bufio.NewReader(r0), |