[Python-Dev] Running Py2.6 with the -3 option

[Brett Cannon]

On Fri, Jul 11, 2008 at 12:26 PM, Adam Olsen <rhamph at gmail.com> wrote:
> On Fri, Jul 11, 2008 at 7:02 AM, Steve Holden <steve at holdenweb.com> wrote:
>> Benjamin Peterson wrote:
>>>
>>> On Fri, Jul 11, 2008 at 6:24 AM, Raymond Hettinger <python at rcn.com> wrote:
>>>>
>>>> Some effort needs to be made to clear the standard library of -3
>>>> warnings.
>>>>  Running -3 on production code usually involves exercising library code
>>>> so
>>>> the useful result is obscured by Python complaining about itself.  Since
>>>> that use case involves the users own tests, I don't think the effort
>>>> needs
>>>> to be extended to our own unittest suite.  But the rest of the library
>>>> could
>>>> likely benefit from a good -3 cleanup.
>>>
>>> Yes, indeed. We should make sure, however, that the changes in the 2.6
>>> libraries are the absolute minimum to get the job done. (I'm trying to
>>> pretend like this isn't violating the prohibition on all-inclusive
>>> overhauls in the stdlib.)
>>>
>> The prohibition is on *gratuitous* changes, basically along the lines of "if
>> it ain't broke, don't fix it". The stdlib is definitely broken if it raises
>> warnings of that kind.
>
> Is the stdlib broken or is it the warnings that are broken?

Nothing is broken, per se, but the stdlib emits a ton of warnings
through basic usage for Py3K-related changes. We are telling people to
run their code in 2.6 with -3 and to eliminate all warnings in order
to have 2to3 work to transition to 3.0. Having the stdlib itself emit
warnings is just not reasonable.

>  The code
> is just fine in 2.6.  Adding pragmas to disable warnings would be just
> fine.  Or we could hardcode some warnings as "already seen".
>

No, we should eat our own dog food and transition the code over. If
anything it will help with code maintenance between 2.x and 3.x.

-Brett