MainInclusionReport for libproxy

MIR team, please see thread at http://lists.debian.org/debian-devel/2008/12/msg00730.html.

We really should disable WPAD by default in our package.

Promoted to main since this blocks GNOME in jaunty, as well as testing the gvfs/samba fix we intend to backport to Hardy (which should happen RSN). Still needs WPAD fix.

Packaging is fine otherwise.

I think it's ok to promote ASAP to unblock GNOME, but found some issues worth fixing:
10:13 <@lool> pochu: libproxy-dev should be arch: any; it doesn't make a
              substantial space saving to have it arch: all and will cause
              installability issues on slow arches
10:14 <@lool> pochu: python-libproxy is arch: all but has a depends on
              shlibs:Depends?
10:15 <@lool> No AM_MAINTAINER_MODE and the configure patch is in the wrong
              order; will cause an autofoo update

I had read the thread on libproxy on debian-devel@, but I re-read it; the conclusion is that WPAD is disabled by default by gnome-network-properties, but might is enabled by default outside of GNOME. http://mail.gnome.org/archives/desktop-devel-list/2008-December/msg00160.html
Critic of WPAD's flaws: http://www.mercenary.net/blog/index.php?/archives/42-HOWTO-WPAD.html

Did proposed cleanups and others and added a patch to disable WPAD fall back by default (with an env var override) in Debian's packaging SVN; would like Emilio to test.

Loïc Minier wrote:
> Did proposed cleanups and others and added a patch to disable WPAD fall
> back by default (with an env var override) in Debian's packaging SVN;
> would like Emilio to test.

If asked upstream to look at
http://code.google.com/p/libproxy/issues/detail?id=20 and he told me he would
try to look at it this week.

Emilio, marking you as the responsible for this bug for jaunty; if this is incorrect for any reason, please let us know.

Steve Langasek wrote:
> Emilio, marking you as the responsible for this bug for jaunty; if this
> is incorrect for any reason, please let us know.

I just looked at Loïc's patch, fixed and tested it, and it's being uploaded to
Debian. If it doesn't get out of NEW soon (say 1 week) I'll get it sponsored,
otherwise I'll request a sync for it.

This bug was fixed in the package libproxy - 0.2.3-0ubuntu4

---------------
libproxy (0.2.3-0ubuntu4) jaunty; urgency=low

  * Upload package from Debian NEW to Jaunty.
  * 50_px-wpad-fallback-env-var disables WPAD fallback (lp: #314945).

 -- Emilio Pozuelo Monfort <email address hidden> Tue, 10 Mar 2009 22:49:57 +0100

We got reports upstream of libproxy: libproxy is not working on Ubuntu.
A user, even when having a Gnome configuration set, get's a warning wpad disabled and the lib falls back to direct, bypassing his proxy.

I would say this part of your 50_px-wpad-fallback-env-var.patch is wrong:
@@ -591,7 +608,12 @@ px_proxy_factory_get_proxies (pxProxyFac
  }

  // If the config plugin returned an invalid config type or malformed URL, fall back to 'wpad://'
- if (!(!strncmp(config->url, "http://", 7) ||
+ if (!do_wpad_fallback)
+ {
+ fprintf(stderr, "*** Config plugin returned invalid URL type and WPAD fallback disabled! Falling back to direct...\n");
+ goto do_return;
+ }
+ else if (!(!strncmp(config->url, "http://", 7) ||
     !strncmp(config->url, "socks://", 8) ||
     !strncmp(config->url, "pac+", 4) ||
     !strcmp (config->url, "wpad://") ||

The only check here seems to be if do_wpad_fallback is enabled, if not give an error and fall back to direct://, without checkling if the config otherwise would have been valid.

I suggest this to && linked with the rest of the original IF.

This is the main inclusion report and was processed. Can you please open a new report about the bug and subscribe pochu?