Why does url-retrieve ignore url-mime-accept-string and most other url variables?

["Clément Pit--Claudel" <clement.pit@gmail.com>]

[-- Attachment #1.1: Type: text/plain, Size: 1276 bytes --]

The doc string of url-retrieve says:

> The variables ‘url-request-data’, ‘url-request-method’ and
> ‘url-request-extra-headers’ can be dynamically bound around the
> request; dynamic binding of other variables doesn’t necessarily
> take effect.

What's the proper way, then, to set ‘url-mime-encoding-string’, ‘url-mime-charset-string’, and ‘url-mime-accept-string’ for a single query? In particular, how come things work with ‘url-retrieve-synchronously’, but not with ‘url-retrieve’?

More precisely: this sends a query with ‘Accept: */*’, and prints nil:

    (let ((url-mime-accept-string "abc/def"))
      (url-retrieve "http://www.gnu.org/software/emacs/"
                    (lambda (&rest _)
                      (message "url-retrieve: %S"
                               url-mime-accept-string))))

but this sends a query with ‘Accept: abc/def’, and prints “abc/def”:

    (let ((url-mime-accept-string "abc/def"))
      (with-current-buffer
          (url-retrieve-synchronously
           "http://www.gnu.org/software/emacs/")
        (message "url-retrieve: %S"
                 url-mime-accept-string)))

Is that a bug? If not, could we document the proper way to set an Accept header?


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]