pylint or similar to test version-specific language constructs?

[Chris Angelico]

On Mon, Jan 14, 2013 at 4:56 AM, jkn <jkn_gg at nicorp.f9.co.uk> wrote:
> Hi Dave
>
> On 11 Jan, 15:06, Dave Angel <d... at davea.name> wrote:
>>
>> Not sure what you mean by beforehand.  Don't you run all your unit tests
>> before putting each revision of your code into production?  So run those
>> tests twice, once on 2.7, and once on 2.4.  A unit test that's testing
>> code with a ternary operator will fail, without any need for a separate
>> test.
>>
>> if it doesn't, then you've got some coverage gaps in your unit tests.
>
> By 'beforehand' I meant 'before testing on my target 2.4 system;
> perhaps I should have been clearer in that I am running 2.7 on my
> 'development' platform, and 2.4 on my target. It would be painful to
> put 2.4 on my target system (although I continue to wonder about
> that...). So I was looking to catch such errors before migrating to
> the target.

Painful to put 2.4 on your dev, you mean? I've never done it, but I
would expect that the old sources will compile against newer libraries
with no problems.

That's likely to be the easiest option. It's the language-level
equivalent of watching for a thrown exception rather than asking
forgiveness beforehand :)

ChrisA