The question is, if we get rid of version numbers in the folder names, will it break anything we care about?

Daniel Holth kirjoitti 01.09.2017 klo 23:38:
A long time ago pkg_resources was able to sometimes get the version number only from the .egg-info folder name, and skip opening the metadata at all when that data was not needed. We might always do listdir() for path in sys.path and parse every METADATA on the first pkg_resources import.

On Fri, Sep 1, 2017 at 4:32 PM Alex Grönholm <alex.gronh...@nextday.fi <mailto:alex.gronh...@nextday.fi>> wrote:

    +1 for getting rid of version numbers in the dist-info folders.


    RonnyPfannschmidt kirjoitti 01.09.2017 klo 20:32:
    > Hi everyone,
    >
    > a while now i have wondered - why put version numbers into the
    dist info
    > folders
    > not only makes it lookup more expensive (need to search for a
    > distribution->folder map)
    > its also not serving a purpose, as the rest of the package is
    > unversioned at the root of site-packages.
    >
    > in addition all those versions in folder names, that serve no
    purpose,
    > and create duplication/conflicts now creep into other peps as well,
    >
    > please lets take a note at setuptools - for editable installs it
    wont
    > put a version in to the egg info folder
    >
    > and imho thats the correct way to deal with data thats not actually
    > versioned at that level where its being stored.
    >
    >
    > -- Ronny
    >
    > _______________________________________________
    > Distutils-SIG maillist  - Distutils-SIG@python.org
    <mailto:Distutils-SIG@python.org>
    > https://mail.python.org/mailman/listinfo/distutils-sig

    _______________________________________________
    Distutils-SIG maillist  - Distutils-SIG@python.org
    <mailto:Distutils-SIG@python.org>
    https://mail.python.org/mailman/listinfo/distutils-sig


_______________________________________________
Distutils-SIG maillist  -  Distutils-SIG@python.org
https://mail.python.org/mailman/listinfo/distutils-sig

Reply via email to