Accept web3 1.2.0 as valid version

[cgewecke]

Hi, am seeing a handlful of CI failures today in projects that float their truffle version and use the zeppelin helpers. Their latest release (v5.0.29) upgrades to the new Web3 stable (v1.2.0) and triggers this error:

Error: web3@1.2.0 detected, incompatible with requirement of web3@1.0.0-beta.37
    at setWeb3 (/home/circleci/project/openzeppelin-solidity/node_modules/openzeppelin-test-helpers/src/configure-web3.js:18:11)
    at defaultConfigure (/home/circleci/project/openzeppelin-solidity/node_modules/openzeppelin-test-helpers/configure.js:48:3)

Examples
gnosis-hg greenkeeper CI
solidity-coverage nightly CI

Additional context about semver.satisfies

semver.satisfies('2.0.0-alpha', '>=1.2.0')
> false
semver.satisfies('2.0.0-alpha.1', '>=1.2.0')
> false
semver.satisfies('1.2.1', '>=1.2.0')
> true
semver.satisfies('2.0.0', '>=1.2.0')
> true

Another possibility is 1.0.0-beta.37 || 1.x but there are some (wrongly) published versions below 1.2....

semver.satisfies('1.2.1', '1.0.0-beta.37 || 1.x')
> true
semver.satisfies('1.0.0-beta.36', '1.0.0-beta.37 || 1.x')
> false
semver.satisfies('2.0.0-alpha', '1.0.0-beta.37 || 1.x')}
> false
semver.satisfies('2.0.0-alpha.1', '1.0.0-beta.37 || 1.x')
> false
semver.satisfies('2.0.0', '1.0.0-beta.37 || 1.x')
> false

Fixes #62

[nventuro]

Hey there @cgewecke, thank you very much for tackling this! We've just got a report about this from @ccolorado in #62, but this seems to be a bit more widespread than I anticipated.

I'll thrown an integration test to make sure this stays fixed, and publish the changes as v0.4.1.

semver.satisfies('2.0.0', '>=1.2.0')
> true

I worry this may cause issues later down the road, perhaps it'd be best to use ^ instead of >=?

Also, just out of curiosity, how did you stumble upon these CI failures? Are you running some sort of monitoring tool, or was it simply a coincidence?

[cgewecke]

was it simply a coincidence

Yes - I'm just watching them on GH and got notifications. Solidity-coverage is my CI job and Gnosis installed a tool I'm writing that I want to make sure works ok.

[cgewecke]

Ah! I was looking at the semver docs but didn't see the caret ranges - will fix rn.

[nventuro]

I ended up creating a sort of 'template' test from the old truffle integration test, which is now reused on both the 5.0.1 and 5.0.29 tests.

@frangio could you take a quick look at this so we can release a bugfix?

[frangio]

Thanks @cgewecke!

@nventuro I left a few comments about the new integration test setup.

[frangio]

-i with no arguments is not portable, it doesn't work in macOS. (However, see next comment.)

Suggested change

	sed -i "s/\"truffle\": \"5.0.1\"/\"truffle\": \"5.0.29\"/g" simple-project-truffle-template/package.json
	sed -i.bak "s/\"truffle\": \"5.0.1\"/\"truffle\": \"5.0.29\"/g" simple-project-truffle-template/package.json
	rm simple-project-truffle-template/package.json.bak

[frangio]

I really think it would be best to simply use npm install.

Suggested change

	sed -i "s/\"truffle\": \"5.0.1\"/\"truffle\": \"5.0.29\"/g" simple-project-truffle-template/package.json
	cd simple-project-truffle-template
	npm install truffle@5.0.29

[frangio]

If the above suggestion is accepted.

Suggested change

	./simple-project-truffle-template/run.sh
	./run.sh

[nventuro]

The npm install approach is much better than my sed hack, yes. @cgewecke would you mind accepting those two last suggestions so we can marge this and release? Thanks!

[cgewecke]

@nventuro Yes sounds good! Please feel free to modify this PR however you wish. :)