Re: Help understanding the URL code

[Lars Magne Ingebrigtsen]

Clément Pit--Claudel <address@hidden> writes:

> Thus IIUC this bit of code will return different values depending on
> whether the url package had already been loaded before or whether it
> was loaded through url-retrieve-synchronously being autoloaded. Is
> that right?

Yes, you're right, and that doesn't make much sense.

Your patch looks good, but perhaps this part could be changed a bit:

[...]

> +    (format "User-Agent: %s\r\n"
> +            (string-trim
> +             (cond
> +              ((functionp url-user-agent) (funcall url-user-agent))
> +              ((stringp url-user-agent) url-user-agent)
> +              (t (url-http--user-agent-default-string)))))))

It should be possible to not have a User-Agent header at all.  nil as a
value to `url-user-agent' would be the obvious choice, and `default'
could be the value to signal using `url-http--user-agent-default-string'.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no