compress/gzip: possible performance regression vs 1.0.3

[davecheney]

What steps will reproduce the problem?
1. git clone https://github.com/davecheney/autobench
2. cd autobench
3. make bench

# autobench is a small harness for running the Go 1.1 go1 tests against go 1.0.3 and tip.

What is the expected output? What do you see instead?

On linux/amd64 several people have reported Gzip encode has increased slightly

BenchmarkGzip                     472108600    486750544   +3.10%

Please use labels and text to provide additional information.

[davecheney]

Comment 1:

No performance regression is visible in the same proportion on 386 or arm (see sample
results in the autobench repository)

[robpike]

Comment 2:

Labels changed: added go1.2maybe, removed go1.1maybe.

[davecheney]

Comment 3:

Labels changed: added priority-later, performance, removed priority-triage.

[rsc]

Comment 4:

Labels changed: added feature.

[robpike]

Comment 5:

Not worth worrying about. Reopen if you disagree.

Status changed to WontFix.