code review 8173043: bytes: don't compact Buffer so aggressively

bytes: don't compact Buffer so aggressively

benchmark                           old ns/op    new ns/op    delta
BenchmarkBufferNotEmptyWriteRead       848416       819983   -3.35%

Update Issue 5154

[bradfitz, 2013-03-29 19:58:01]

Hello golang-dev@googlegroups.com,

I'd like you to review this change to
https://go.googlecode.com/hg/

[gri, 2013-03-29 20:37:39]

LGTM

https://codereview.appspot.com/8173043/diff/2001/src/pkg/bytes/buffer.go
File src/pkg/bytes/buffer.go (right):

https://codereview.appspot.com/8173043/diff/2001/src/pkg/bytes/buffer.go#newc...
src/pkg/bytes/buffer.go:91: // We can slide things down instead of
Please mention why the choice of 1/2 the buffer in the comment. Not immediately
evident.

https://codereview.appspot.com/8173043/diff/2001/src/pkg/bytes/buffer_test.go
File src/pkg/bytes/buffer_test.go (right):

https://codereview.appspot.com/8173043/diff/2001/src/pkg/bytes/buffer_test.go...
src/pkg/bytes/buffer_test.go:493: if cap1 > cap0*3 {
Please add a comment regarding the choice of 3. This will be somewhat mysterious
in 1/2 year...

[robryk, 2013-03-29 20:39:07]

On 2013/03/29 19:58:01, bradfitz wrote:
> Hello mailto:golang-dev@googlegroups.com,
> 
> I'd like you to review this change to
> https://go.googlecode.com/hg/

I've written a benchmark that verifies that this helps:
https://codereview.appspot.com/8174043/

If you want me to mail it, just say so.

[bradfitz, 2013-03-29 20:49:18]

*** Submitted as https://code.google.com/p/go/source/detail?r=92913ac2c60f ***

bytes: don't compact Buffer so aggressively

benchmark                           old ns/op    new ns/op    delta
BenchmarkBufferNotEmptyWriteRead       848416       819983   -3.35%

Update Issue 5154

R=golang-dev, gri, robryk
CC=golang-dev
https://codereview.appspot.com/8173043

[bradfitz, 2013-03-29 20:49:21]

Done.

On Fri, Mar 29, 2013 at 1:37 PM, <gri@golang.org> wrote:

> LGTM
>
>
> https://codereview.appspot.**com/8173043/diff/2001/src/pkg/**
>
bytes/buffer.go<https://codereview.appspot.com/8173043/diff/2001/src/pkg/bytes/buffer.go>
> File src/pkg/bytes/buffer.go (right):
>
> https://codereview.appspot.**com/8173043/diff/2001/src/pkg/**
>
bytes/buffer.go#newcode91<https://codereview.appspot.com/8173043/diff/2001/src/pkg/bytes/buffer.go#newcode91>
> src/pkg/bytes/buffer.go:91: // We can slide things down instead of
> Please mention why the choice of 1/2 the buffer in the comment. Not
> immediately evident.
>
> https://codereview.appspot.**com/8173043/diff/2001/src/pkg/**
>
bytes/buffer_test.go<https://codereview.appspot.com/8173043/diff/2001/src/pkg/bytes/buffer_test.go>
> File src/pkg/bytes/buffer_test.go (right):
>
> https://codereview.appspot.**com/8173043/diff/2001/src/pkg/**
>
bytes/buffer_test.go#**newcode493<https://codereview.appspot.com/8173043/diff/2001/src/pkg/bytes/buffer_test.go#newcode493>
> src/pkg/bytes/buffer_test.go:**493: if cap1 > cap0*3 {
> Please add a comment regarding the choice of 3. This will be somewhat
> mysterious in 1/2 year...
>
>
https://codereview.appspot.**com/8173043/<https://codereview.appspot.com/8173...
>

[bradfitz, 2013-03-29 20:49:34]

On Fri, Mar 29, 2013 at 1:39 PM, <robryk@gmail.com> wrote:

> On 2013/03/29 19:58:01, bradfitz wrote:
>
>> Hello mailto:golang-dev@**googlegroups.com <golang-dev@googlegroups.com>,
>>
>
>  I'd like you to review this change to
>> https://go.googlecode.com/hg/
>>
>
> I've written a benchmark that verifies that this helps:
>
https://codereview.appspot.**com/8174043/<https://codereview.appspot.com/8174...
>
> If you want me to mail it, just say so.
>

Go for it.