Don't allcoate a new math/rand every request.

Don't allcoate a new math/rand every request.

Use a global math/rand instance, rather than re-allocating one for each
request.

BUG=chromium:675485
TEST=None
R=vadimsh@chromium.org

Review-Url: https://codereview.chromium.org/2594463002
Committed: https://github.com/luci/luci-go/commit/fd0a102c24af05830dccbce7aeb7cf8d3295e90c

[dnj, 2016-12-20 00:50:41]

PTAL. I'm seeing a lot of math/rand allocations in my app's heap, and AFAICT
there's no reason not to use a global one other than a shared lock. However, the
real "math/rand" uses a global locked rand, so it's probably not too bad.

[Vadim Sh., 2016-12-20 00:55:02]

lgtm

Another approach is to keep a sync.Pool of RNG instances and give each request
it's own one (returning it back to the pool when done, without reseting).

[dnj, 2016-12-20 00:56:31]

On 2016/12/20 00:55:02, Vadim Sh. wrote:
> lgtm
> 
> Another approach is to keep a sync.Pool of RNG instances and give each request
> it's own one (returning it back to the pool when done, without reseting).

Is the advantage of this just less lock contention? I think a global one will be
okay; this isn't crypto, so it shouldn't be bound to an actual blocking device.

[dnj, 2016-12-20 00:56:33]

The CQ bit was checked by dnj@chromium.org

[Vadim Sh., 2016-12-20 00:56:37]

https://codereview.chromium.org/2594463002/diff/1/appengine/gaemiddleware/con...
File appengine/gaemiddleware/context.go (right):

https://codereview.chromium.org/2594463002/diff/1/appengine/gaemiddleware/con...
appengine/gaemiddleware/context.go:66: // globalMathRand is a global "math/rand"
instance.
actually, nit: add a comment stating that this object has locks inside

[commit-bot: I haz the power, 2016-12-20 00:56:42]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[dnj, 2016-12-20 00:57:43]

The CQ bit was unchecked by dnj@chromium.org

[dnj, 2016-12-20 00:58:49]

https://codereview.chromium.org/2594463002/diff/1/appengine/gaemiddleware/con...
File appengine/gaemiddleware/context.go (right):

https://codereview.chromium.org/2594463002/diff/1/appengine/gaemiddleware/con...
appengine/gaemiddleware/context.go:66: // globalMathRand is a global "math/rand"
instance.
On 2016/12/20 00:56:37, Vadim Sh. wrote:
> actually, nit: add a comment stating that this object has locks inside

Done.

[dnj, 2016-12-20 00:59:01]

The CQ bit was checked by dnj@chromium.org

[dnj, 2016-12-20 00:59:01]

The patchset sent to the CQ was uploaded after l-g-t-m from vadimsh@chromium.org
Link to the patchset: https://codereview.chromium.org/2594463002/#ps20001
(title: "Lock comment.")

[commit-bot: I haz the power, 2016-12-20 00:59:12]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-12-20 01:06:46]

CQ is committing da patch.

Bot data: {"patchset_id": 20001, "attempt_start_ts": 1482195541187400,
"parent_rev": "fad4dadaa7d4d99fc00958a4d5281af7f9d85a7c", "commit_rev":
"fd0a102c24af05830dccbce7aeb7cf8d3295e90c"}

[commit-bot: I haz the power, 2016-12-20 01:06:49]

Description was changed from

==========
Don't allcoate a new math/rand every request.

Use a global math/rand instance, rather than re-allocating one for each
request.

BUG=chromium:675485
TEST=None
R=vadimsh@chromium.org
==========

to

==========
Don't allcoate a new math/rand every request.

Use a global math/rand instance, rather than re-allocating one for each
request.

BUG=chromium:675485
TEST=None
R=vadimsh@chromium.org

Review-Url: https://codereview.chromium.org/2594463002
Committed:
https://github.com/luci/luci-go/commit/fd0a102c24af05830dccbce7aeb7cf8d3295e90c
==========

[commit-bot: I haz the power, 2016-12-20 01:06:50]

Committed patchset #2 (id:20001) as
https://github.com/luci/luci-go/commit/fd0a102c24af05830dccbce7aeb7cf8d3295e90c