Publisher::WaitForConnection() is being used without timeout

[osrf-migration]

Original report (archived issue) by Steve Peters (Bitbucket: Steven Peters, GitHub: scpeters).

---

I have observed occasional hangs at Publisher::WaitForConnection(), which has a potentially infinite loop. There is an alternative version of the function with takes a common::Time as a timeout argument, but it's only being used by gzlog.cc:959.

The function is called without a timeout in the following places:

• gzfactory.cc:118
• gzfactory.cc:156
• Scene.cc:289
• Transport.cc:212
• Node.hh:54
• test/ServerFixture.hh:174-175
• test/regression/state_log.cc:83
• test/regression/state_log.cc:150
• test/regression/transport.cc:118
• examples/stand_alone/publisher/publisher.cc:45

Seeing Scene.cc in this list makes me think of issue #681 (blank gzclient windows on startup).

[osrf-migration]

Original comment by Steve Peters (Bitbucket: Steven Peters, GitHub: scpeters).

---

• set component to "transport"

[osrf-migration]

Original comment by Steve Peters (Bitbucket: Steven Peters, GitHub: scpeters).

---

Noted another timeout failure while testing INTEGRATION_transceiver while testing pull request #811.

[osrf-migration]

Original comment by Jose Luis Rivero (Bitbucket: Jose Luis Rivero, GitHub: j-rivero).

---

Use fixed values timeouts for testing is a bit tricky since the machine performance could vary, which could end up in false positives. They could be valid when we expect a value and get a result which is clearly above or under the magnitude order we expect. To optimize the check, probably the ideal case would be to overload the Publisher::WaitForConnection() to include a callback.

If we don't set an explicit Timeout here, gtest will kill it after 240 secs right? That should cover the infinite loops and problems with the async calls.

[osrf-migration]

Original comment by Steve Peters (Bitbucket: Steven Peters, GitHub: scpeters).

---

True, gtest will kill it after 240 seconds, but gzserver and gzclient would hang if it happened. I wonder if ignition-transport will have this same problem? @caguero

[osrf-migration]

Original comment by Nate Koenig (Bitbucket: Nathan Koenig).

---

• set version to "all"