Fixes #258, handles undefined Response object rather than throwing an…

[mmerkes]

… uncaught error

I kept getting this error on random occasions:

/Users/Matt/open_source/express-route-validator/node_modules/supertest/lib/test.js:202
    if (res.status !== status) {
           ^
TypeError: Cannot read property 'status' of undefined
    at Test.assert (/Users/Matt/open_source/express-route-validator/node_modules/supertest/lib/test.js:202:12)
    at Server.assert (/Users/Matt/open_source/express-route-validator/node_modules/supertest/lib/test.js:131:12)
    at Server.g (events.js:180:16)
    at Server.emit (events.js:92:17)
    at net.js:1276:10
    at process._tickCallback (node.js:419:13)

When I dug down further, I could set that when res was undefined, it was because the server was unexpectedly timing out:

Error: connect ETIMEDOUT
    at errnoException (net.js:904:11)
    at Object.afterConnect [as oncomplete] (net.js:895:19)

So, if we check for the error first and fire the callback immediately there's not res, the problem is resolved.

This PR is an alternative to #240 and fixes #237

[Starefossen]

@mikelax any eta on when this may get merged?

[mikelax]

I am planning to try and get the 1.1.0 release out next week.

[Starefossen]

Nice, looking forward to that 😄

[mikelax]

@mmerkes I just merged in #239 which was a big refactor to test.js
Do you want to resolve the conflicts for this PR?

[mmerkes]

@mikelax Yeah, I'm happy to resolve the merge conflicts.

[mmerkes]

@mikelax Meant to knock this out earlier, but been crazy around here. It turns out that the refactor resolved the issues around my changes, so my changes in lib/test.js are obsolete, but I kept the tests that were previously failing (without my fix) to confirm the issue was resolved.

[Starefossen]

Thanks for fixing this @mikelax. Could you also tag the v1.1.0 version in Git/GitHub?

[mikelax]

np @Starefossen I just created the v1.1.0 release.

[kadishmal]

I don't see any changes in this PR other than the test case. Should it fix something? It definitely doesn't fix #258.

[mmerkes]

@kadishmal As described in the thread, I had fixed this issue in the 1.0 code, but there was a major refactor of the file in 1.1 that held the root of the issue which resolved the problem. The tests that were added were cases that I was able to repro the issue in 1.0, but they pass with no fix in 1.1; thus, the PR became tests confirming the issue was resolved rather than the resolution. If you're still seeing this in 1.1, you should include the repro.