Re: Embedded six.py in many packages: can someone add a lintian check?
On 28.05.2014 16:02, Barry Warsaw wrote:
> On May 28, 2014, at 05:18 PM, Thomas Goirand wrote:
>
>> To avoid that it grows out of proportion, can someone add a lintian
>> check for that? I'd be *very* happy to have this check for myself as
>> well, so that I don't miss it once more (it's quite easy to miss...).
>
> Unfortunately, it's not just six that gets vendorized. I'd be in favor of a
> lintian check if it could be generalized to warn against all vendorizing. A
> warning specifically about six is helpful but limited.
>
>> Also, am I the only one that believe this is a bug, and that we should
>> do mass bug-filling?
>
> It's definitely a bug.
>
a low importance bug in my opinion.
six is not friendly to avoiding embedding upstream as it does weird
stuff with moving names so the usual __init__.py override does not work.
(I just symlinked the file in the end)
Also it does not seem to have a stable API, I just changed back to
embedded six in python-scipy because what we have in unstable now is not
compatible with what scipy is using now (1.2).
Upstream also has zero incentive to update their embedded version as it
gains them nothing but a larger copied file.
Reply to: