Fixed GAUGE32 integer overflow in base_client.rb

[pprugger]

As discussed here:
#64

Best regards

[pprugger]

I changed the dist to trusty, only way i found to get it building with oraclejdk8 again.

[colinsurprenant]

Thanks a lot @pprugger !

• Could you remove the .travis.yml change, this is really non related to this PR; I will push a new PR shortly which will solve this, you can then rebase to pick it up and tests should pass.
• Would you mind adding a test/spec for this? you could add it into spec/inputs/snmp/base_client_spec.rb in the coercion section. Just add a test with a value that overflows without the patch and works with it. Let me know if you need help for that!

[colinsurprenant]

FYI build fixed in #66

[pprugger]

I hope it's not too much of chaos now, had some problems with rebaseing.
To be honest i have no clue how to add the test.
I would test with uint_max so 4294967295, which should be 4294967295 if it works, otherwise it should be -1. (if the overflow occurs)

[colinsurprenant]

@pprugger thanks for the dist revert. I'll send an example spec which will fail locally for me without patch and works for you, you can then include it in your PR!

[colinsurprenant]

@pprugger here, you can add these 2 tests in `spec/inputs/snmp/base_client_spec.rb" in the "coercion" context:

      it "should handle max unsigned 32 bits integer GAUGE32" do
        MAX_UNSIGNED_INT_32 = 4294967295
        v = Gauge32.new(MAX_UNSIGNED_INT_32)
        expect(subject.coerce(v)).to eq(MAX_UNSIGNED_INT_32)
      end

      it "should handle max signed 32 bits integer INTEGER32" do
        MAX_SIGNED_INT_32 = 2147483647
        v = Integer32.new(MAX_SIGNED_INT_32)
        expect(subject.coerce(v)).to eq(MAX_SIGNED_INT_32)
      end

The first one for GAUGE32 does not pass for me and returns -1 as you noted and should pass for you, the second for INTEGER32 does pass which is normal; I just added it for the sake of completeness.

[pprugger]

Did you change something else? It seems the names Gauge32 and Integer32 are not known in my context.

[pprugger]

I added the java imports. Works now and can be merged if you want.

[colinsurprenant]

cosmetic: the identation is off

[colinsurprenant]

@pprugger great stuff!! yeah, forgot to mention the java imports but you figured it out 👍
All good, minus a small cosmetic nitpick!

[pprugger]

Fixed!

Will you do a new release after the merge?

[colinsurprenant]

Thanks a lot again for your contribution @pprugger. Feel free to submit a PR to add your information into the CONTRIBUTORS file.
I will merge this, bump the version and then publish it shortly.

[colinsurprenant]

v1.2.1 published.