[py3porters-devel] [Needs Audit] Debian Infra Depends

Scott Kitterman debian at kitterman.com
Sat Apr 18 07:25:25 UTC 2015


On April 18, 2015 1:40:19 AM EDT, Daniel Kahn Gillmor <dkg at fifthhorseman.net> wrote:
>On Sat 2015-04-18 04:38:30 +0200, Bryan Shook wrote:
>> So a cursory glance through this list and I spotted a package that
>won't
>> need creation, python-pysqlite2.  That code became the sqlite3
>module.
>
>in the same vein, do we want to try to maintain a new python3-psycopg2
>when python3-postgresql offers a (possibly more modern) python3 DB-API
>driver?

Since "we" is me and I've been doing it for over three years already, I'll go out on a limb and say yes.  "We" do.

>We can do both, of course, but it may be that we want to consider with
>some more nuance about what the right upgrade path is for each py2
>module.
>
>so far, we've seen the following exceptions to the "we want a
>python3-foo package for every python-foo package" rule of thumb:
>
> a) python-foo binary packages that contain py3 code

For modules and extensions this is generally a bug that should be fixed. 

>b) python-foo binaries where the upgrade path to python3 is python3-bar
>    (python3-foo simply won't be supported)
>
>c) python-foo binaries where the upgrade path to python3 is python3-bar
>    (python3-foo is available but deprecated)
>
>are there other common exceptions?

Cases of a) that aren't bugs aren't common. 

Scott K




More information about the py3porters-devel mailing list