Most changes to Python require a NEWS entry.

Please add it using the blurb_it web app or the blurb command-line tool.

All commit authors signed the Contributor License Agreement.

Any news on this? The issue is getting spread on more and more computers and may stuck some people with a .netrc shared by multiple applications.

Someone pointed me to: https://cgit.git.savannah.gnu.org/cgit/inetutils.git/tree/ftp/ruserpass.c#n362; the backslash is skipped there as well. This must have to do with the format of netrc files?

The current code matches what is done in GNU inetutils, and supposedly (chatgpt) GNU wget, Curl, and the other ftp implementations for parsing netrc files. Maybe the status quo is better, to preserve compatibility and behavior across tooling?

As an extra data point, I checked what Emacs does and it differs from inetutlis: it doesn't do any unescaping (preserves everything literally), and only strip quotes at the edge of a netrc token:

(defun auth-source-netrc-looking-at-token ()
  "Say whether the next think in the buffer is a token (password, etc).
Match data is altered to reflect the token."
  (or (looking-at "'\\([^']*\\)'")
      (looking-at "\"\\([^\"]*\\)\"")
      (looking-at "\\([^ \t\n]+\\)")))

See: https://cgit.git.savannah.gnu.org/cgit/emacs.git/tree/lisp/auth-source.el#n1040

Hi guys, what's the status here?

Curl does not have any issue treating my password's backslash as a backslash, instead of silently stripping it like the netrc python lib does. I spent two hours yesterday to find out why my python netrc test ran into a 401 authentication failure. Now I wrote my own code instead.

This PR is stale because it has been open for 30 days with no activity.