Python 2.6: URLs should be stripped of #egg= fragments

[ghost]

Originally reported by: rassie (Bitbucket: rassie, GitHub: rassie)

---

Merged setuptools is missing a crucial conditional from distribute in the package_index._download_to function:

if '#' in url:
    url, info = url.split('#', 1)

Because of this, the #egg= is sent to the server and probably URL-encoded. At least for Bitbucket URLs this leads to 404 errors, but it wrong in general, since #egg= is metadata for setuptools only.

This error is probably not visible in Python 2.7+, since urllib2 has been patched to ignore the URL fragment on request.

---

• Bitbucket: https://bitbucket.org/pypa/setuptools/issue/69

[ghost]

Original comment by jaraco (Bitbucket: jaraco, GitHub: jaraco):

---

This issue stems from this changeset (4e4ccf3ea04d), which was ancestral to 0.9, so 0.7 and 0.8 should not be affected by this issue. I remember testing this behavior when I made the change (to ensure that the fragment in the request was suitably ignored), but I did not test on older Python versions. Thanks for the report; I'll get it fixed straight away.

[ghost]

Original comment by jaraco (Bitbucket: jaraco, GitHub: jaraco):

---

Correct 404 errors when URLs contain fragments. Fixes #69.

[ghost]

Original comment by jaraco (Bitbucket: jaraco, GitHub: jaraco):

---

Fixed in 1.1.3

[ghost]

Original comment by jaraco (Bitbucket: jaraco, GitHub: jaraco):

---

Looks like I got the logic wrong in the compat module (bb29cec316de). Should be fixed in 1.1.5.