3
97%
master: 97%

Ran 03 Nov 2014 05:27PM UTC

Files 55

Run time 39s

Badge

Embed ▾

Committed 03 Nov 2014 05:19PM UTC coverage: 95.609% (+0.005%) from 95.604%

Job # 2.1.0, CI="travis"

Build Type

Pull #331

travis-ci

Committed by

telent

Commit Message

socket timeouts using select not SO_RCVTIMEO

In MRI 2.1.3 - and probably most other versions - setting the
SO_RCVTIMEO option on a socket will cause read(2) to return EWOULDBLOCK
as expected, but the interpreter internals will retry the operation
instead of passing the failure up to interpreted code.  Thus, the
timeout has no visible effect.

What we should do instead is call Kernel.select with the required
timeout to check if there is data available, *then* call read(2).
If .select fails we raise a Errors::ConnectionFailure on the assumption
that something higher up in the stack will catch it and cope.

Pull Request Pull Request #331: Hangs forever in refresh after 5 minutes when 2ary unreachable

Run Details

1459 of 1526 relevant lines covered (95.61%)

278.32 hits per line

mongoid / moped / 1112 / 3
97%
master: 97%

README BADGES
x

Markdown

Textile

RDoc

HTML

Rst

Source Files on job 1112.3 (2.1.0, CI="travis")

mongoid / moped / 1112 / 3 97% master: 97%

README BADGES x

Markdown

Textile

RDoc

HTML

Rst

Source Files on job 1112.3 (2.1.0, CI="travis")

mongoid / moped / 1112 / 3
97%
master: 97%

README BADGES
x