[mythtv] Ticket #9525: [PATCH] Mythbrowser does not use proxy set in http_proxy environment variable

[Michael T. Dean]

On 05/09/2011 03:44 AM, MythTV wrote:
> #9525: [PATCH] Mythbrowser does not use proxy set in http_proxy environment
> variable
> ---------------------------------------+----------------------------
>   Reporter:  jan.ceuleers@…             |          Owner:  paulh
>       Type:  Patch - Bug Fix            |         Status:  new
>   Priority:  minor                      |      Milestone:  unknown
> Component:  Plugin - MythBrowser       |        Version:  0.23-fixes
>   Severity:  low                        |     Resolution:
>   Keywords:                             |  Ticket locked:  0
> ---------------------------------------+----------------------------
>
> Comment (by nigel@…):
>
>   1. I wanted to add proxy support for CDDB lookups many years ago, but that
>   required a different way to do the lookups. When (if?) !HttpComms is used
>   for that, '''''it''''' should be the location for doing proxy-compatible
>   stuff?
>   2. My gut feeling is that using an environment variable is a little clumsy
>   - I would have preferred a setting from the database, but then that is
>   just another thing that needs a settings UI.
>   3. Mac OS X and Windows supposedly have hooks to look up the system proxy
>   settings via QNetworkProxyFactory::systemProxyForQuery(). I find it hard
>   to believe that the Linux environments don't have a standard proxy setting
>   location these days?
>   4. Now that the backend also does HTTP traffic, it also needs to set proxy
>   "environment"s?
>   Paul, my work has a few corporate proxies - I'm happy to take this is you
>   don't have any burning desire.

Nigel,

FWIW, my (non-proxy-using) understanding is that the generally-accepted 
de facto "standard" approach for configuring proxy support in *nix is 
the use of the http_proxy and https_proxy environment variables.  So, 
using this--rather than a DB setting that the user has to find--would 
make our application work the same way as a lot of other *nix software.  
Chances are, users who do have proxies, already have the environment 
variable configured for some other software, anyway.

http://www.google.com/search?hl=en&q=http_proxy
http://www.google.com/search?hl=en&q=http_proxy+linux

See, also, http://bugreports.qt.nokia.com/browse/QTBUG-14623 .  Qt devs 
are looking at supporting a *nix implementation for 
QNetworkProxyFactory::setUseSystemConfiguration() (possibly by 4.8.0), 
but  are trying to decide whether to include just http*_proxy 
environment variables, or also to use GNOME/KDE configuration.

Personally (OK, this may be bias due to my, "We have way too many 
settings in MythTV," stance), I'd rather just use the system-configured 
proxy settings rather than create our own setting.  If that means using 
http*_proxy and GNOME's (~/.gconf/system/http_proxy and 
~/.gconf/system/proxy) and KDE's (no idea where they keep theirs) and 
xdg-settings proxy configuration ( 
http://www.crystalorb.net/mikem/xdg-settings.html --which, admittedly, 
isn't finished, yet), that approach would be fine with me.  
However--especially if Qt is going to include support for more-specific 
settings in the future--starting with the http*_proxy environment 
variables seems reasonable for now.  And if anyone wants to submit 
patches that determine proxy settings using GNOME/KDE/xdg configuration, 
that would work, too (though I'd argue that it's more useful to submit 
those patches to Qt on the bug report I referenced, above).

Mike