Bug #875266 “Firefox ignores GNOME 3 proxy settings” : Bugs : Mozilla Firefox

Revision history for this message

In Mozilla Bugzilla #682832, Jhorak (jhorak) wrote on 2011-08-29:

#2

Gnome 3 uses GSettings for proxy configuration now. Firefox does not get proxy settings set by gnome-control-center when option 'Use system proxy settings' in Connection settings is set.

Revision history for this message

In Mozilla Bugzilla #682832, Jhorak (jhorak) wrote on 2011-08-29:

#3

Created attachment 556542
gsettings proxy patch v 1

Initial patch, not sure who to set as reviewer, feel free to reassign.

Revision history for this message

In Mozilla Bugzilla #682832, Karlt (karlt) wrote on 2011-09-05:

#4

Is this setting fetched during start-up?

I'm worried about using GSettings during start-up. (Bug 611953 comment 37)
Using GSettings during a delayed request after the browser is visible would be acceptable, I hope.

A stack track at the first invocation of this code may tell the story.

Revision history for this message

Nobuto Murata (nobuto) wrote on 2011-10-15:

#1

AlsaDevices.txt Edit (467 bytes, text/plain; charset="utf-8")
BootDmesg.txt Edit (60.8 KiB, text/plain; charset="utf-8")
Card0.Amixer.values.txt Edit (2.5 KiB, text/plain; charset="utf-8")
Card0.Codecs.codec.0.txt Edit (9.7 KiB, text/plain; charset="utf-8")
Card0.Codecs.codec.1.txt Edit (155 bytes, text/plain; charset="utf-8")
CurrentDmesg.txt Edit (110.6 KiB, text/plain; charset="utf-8")
DefaultProfileExtensions.txt Edit (1.2 KiB, text/plain; charset="utf-8")
DefaultProfileLocales.txt Edit (385 bytes, text/plain; charset="utf-8")
DefaultProfilePlugins.txt Edit (762 bytes, text/plain; charset="utf-8")
DefaultProfilePrefs.txt Edit (7.1 KiB, text/plain; charset="utf-8")
DefaultProfileThemes.txt Edit (129 bytes, text/plain; charset="utf-8")
Dependencies.txt Edit (2.7 KiB, text/plain; charset="utf-8")
IpAddr.txt Edit (806 bytes, text/plain; charset="utf-8")
IwConfig.txt Edit (537 bytes, text/plain; charset="utf-8")
Lspci.txt Edit (13.9 KiB, text/plain; charset="utf-8")
PciMultimedia.txt Edit (598 bytes, text/plain; charset="utf-8")
PciNetwork.txt Edit (1.2 KiB, text/plain; charset="utf-8")
ProcCpuinfo.txt Edit (1.6 KiB, text/plain; charset="utf-8")
ProcEnviron.txt Edit (156 bytes, text/plain; charset="utf-8")
Profile1Extensions.txt Edit (516 bytes, text/plain; charset="utf-8")
Profile1Locales.txt Edit (385 bytes, text/plain; charset="utf-8")
Profile1Plugins.txt Edit (762 bytes, text/plain; charset="utf-8")
Profile1Prefs.txt Edit (820 bytes, text/plain; charset="utf-8")
Profile1Themes.txt Edit (129 bytes, text/plain; charset="utf-8")
Profile2Extensions.txt Edit (637 bytes, text/plain; charset="utf-8")
Profile2Locales.txt Edit (385 bytes, text/plain; charset="utf-8")
Profile2Plugins.txt Edit (915 bytes, text/plain; charset="utf-8")
Profile2Prefs.txt Edit (736 bytes, text/plain; charset="utf-8")
Profile2Themes.txt Edit (129 bytes, text/plain; charset="utf-8")
Profile3Extensions.txt Edit (607 bytes, text/plain; charset="utf-8")
Profile3Locales.txt Edit (367 bytes, text/plain; charset="utf-8")
Profile3Plugins.txt Edit (982 bytes, text/plain; charset="utf-8")
Profile3Prefs.txt Edit (853 bytes, text/plain; charset="utf-8")
Profile3Themes.txt Edit (123 bytes, text/plain; charset="utf-8")
Profile4Extensions.txt Edit (934 bytes, text/plain; charset="utf-8")
Profile4Locales.txt Edit (368 bytes, text/plain; charset="utf-8")
Profile4Plugins.txt Edit (984 bytes, text/plain; charset="utf-8")
Profile4Prefs.txt Edit (1.0 KiB, text/plain; charset="utf-8")
Profile4Themes.txt Edit (123 bytes, text/plain; charset="utf-8")
PulseSinks.txt Edit (2.1 KiB, text/plain; charset="utf-8")
PulseSources.txt Edit (3.6 KiB, text/plain; charset="utf-8")
RfKill.txt Edit (112 bytes, text/plain; charset="utf-8")
WifiSyslog.txt Edit (777.7 KiB, text/plain; charset="utf-8")

Bug Watch Updater (bug-watch-updater) on 2011-10-15

Changed in firefox:
importance:	Unknown → Medium
status:	Unknown → In Progress

Chris Coulson (chrisccoulson) on 2011-10-15

Changed in firefox (Ubuntu):
status:	New → Triaged
importance:	Undecided → Medium

Revision history for this message

ThierryM (thierry-munoz) wrote on 2011-10-17:

#5

Hi,
I confirm the proxy settings aren't applied for Firefox and Thunderbird. But there are applied for Chromium.

Revision history for this message

CuteChaps (sh-senthilkumar) wrote on 2011-10-18:

#6

Firefox/pidgin/thunderbird doesn't pick the proxy settings from system proxy settings.

Revision history for this message

CuteChaps (sh-senthilkumar) wrote on 2011-10-18:

#7

It was working fine on 11.04, multiple users are affected so please help us with a fix or workaround at the earliest.

Revision history for this message

In Mozilla Bugzilla #682832, Jhorak (jhorak) wrote on 2011-10-18:

#8

Created attachment 567719
backtrace first run

Attaching backtrace of first run (when previous version of Firefox was executed last time).

Revision history for this message

In Mozilla Bugzilla #682832, Jhorak (jhorak) wrote on 2011-10-18:

#9

Created attachment 567720
backtrace next run

Revision history for this message

In Mozilla Bugzilla #682832, Jhorak (jhorak) wrote on 2011-10-18:

#10

Created attachment 567725
DumpJSStack

By checking js stacktrace getting proxy settings for this kind of URI makes sense. I can't imagine how to avoid it during startup.

Revision history for this message

In Mozilla Bugzilla #682832, Karlt (karlt) wrote on 2011-10-25:

#11

Thank you. I'm OK with checking settings during startup if it's only for the whatsnew page when there is a new version.

Looks like starting the proxy service is lazy intentionally:
http://hg.mozilla.org/mozilla-central/annotate/767693e248aa/netwerk/base/src/nsIOService.cpp#l653

I'll look further into your patch, thanks.

Revision history for this message

In Mozilla Bugzilla #682832, Karlt (karlt) wrote on 2011-11-25:

#12

Comment on attachment 556542
gsettings proxy patch v 1

I looked at the GSettingsService changes.

There is a new symbol that needs to be added to the GSETTINGS_FUNCTIONS at the
top of the file for dynamic lookup, and some other things to touch up there so
that this compiles against old versions of GIO.

>+#include "nsISupports.h"
> #include "nsCOMPtr.h"
>+#include "nsIMutableArray.h"

I expect nsISupports.h does not need to be listed explicitly because
nsIMutableArray.h or nsISupportsPrimitives.h will pull it in.

>+ const gchar ** gs_strings = g_variant_get_strv(value, NULL);
>+ g_variant_unref(value);

gs_strings doesn't own the strings, only the list of pointers.
What owns the strings after value is released?

>+ if (!gs_strings) {
>+ // empty array
>+ return NS_OK;
>+ }

aResult needs to be set if returning NS_OK.

>+ nsCOMPtr<nsIMutableArray> items(do_CreateInstance(NS_ARRAY_CONTRACTID));
>+ if (!items)
>+ return NS_ERROR_OUT_OF_MEMORY;

Leaks gs_strings.

Probably tidiest to move this construction to before other allocations.

>+ nsCOMPtr<nsISupportsString> obj(do_CreateInstance(NS_SUPPORTS_STRING_CONTRACTID));

Would it make sense to instead use nsISupportsCString?
That would be more consistent with the AUTF8Strings returned by getString and
used in parameters, and it is probably the format more useful to the caller in
this case.

Think about whether using NS_NewAdoptingUTF8StringEnumerator instead of
nsIMutableArray would simplify things:
http://hg.mozilla.org/mozilla-central/annotate/84117219ded0/xpcom/ds/nsStringEnumerator.h#l90

Revision history for this message

In Mozilla Bugzilla #682832, Jhorak (jhorak) wrote on 2011-11-28:

#13

(In reply to Karl Tomlinson (:karlt) from comment #7)
> Think about whether using NS_NewAdoptingUTF8StringEnumerator instead of
> nsIMutableArray would simplify things:
> http://hg.mozilla.org/mozilla-central/annotate/84117219ded0/xpcom/ds/
> nsStringEnumerator.h#l90

Isn't NewAdoptingUTF8StringEnumerator available only for internal API? I'm having problem with compilation when file including nsStringEnumerator.h (it seems to include nsString.h which needs to have MOZILLA_INTERNAL_API defined and I don't see MOZILLA_INTERNAL_API in Makefile or in compilation output - in toolkit/system/gnome directory).

Revision history for this message

In Mozilla Bugzilla #682832, Karlt (karlt) wrote on 2011-11-29:

#14

(In reply to jhorak from comment #8)
> Isn't NewAdoptingUTF8StringEnumerator available only for internal API?

Looks like you are right, based on bug 466622 comment 16 and 17.

Revision history for this message

In Mozilla Bugzilla #682832, Jhorak (jhorak) wrote on 2011-11-29:

#15

Created attachment 577534
gsettings proxy patch v2

Thanks for reviewing. Attaching patch which should respect mentioned issues.

>There is a new symbol that needs to be added to the GSETTINGS_FUNCTIONS at the
>top of the file for dynamic lookup, and some other things to touch up there so
>that this compiles against old versions of GIO.
Could you be more specific (the other things)?

Please have a look.

Revision history for this message

In Mozilla Bugzilla #682832, Karlt (karlt) wrote on 2011-11-30:

#16

Download full text (3.7 KiB)

Comment on attachment 577534
gsettings proxy patch v2

> FUNC(g_variant_get_string, const char *, (GVariant* value, gsize* length)) \
> FUNC(g_variant_is_of_type, gboolean, (GVariant* value, const GVariantType* type)) \
> FUNC(g_variant_new_int32, GVariant *, (gint32 value)) \
> FUNC(g_variant_new_boolean, GVariant *, (gboolean value)) \
> FUNC(g_variant_new_string, GVariant *, (const char* string)) \
>+ FUNC(g_variant_get_strv, GVariant *, (gsize* length)) \

These were roughly in an order.
Can you insert get_strv after get_string, please?

(In reply to jhorak from comment #10)
> Could you be more specific (the other things)?

Check the #define statements before and after GSETTIINGS_FUNCTIONS as any new
symbols need to be added here.

A #define is needed for g_variant_get_strv and G_VARIANT_TYPE_STRING_ARRAY.

>+ if (mGConf && IsProxyMode("auto")) {
>+ return mGConf->GetString(NS_LITERAL_CSTRING("/system/proxy/autoconfig_url"),
>+ aResult);
> }

>+ if (mGSettings) {
>+ // Check if mode is auto

I assume GSettings should override GConf settings.
Otherwise I assume those who upgrade to GNOME 3 will still be using their old
GConf settings, but the configuration utility will change the GSettings
values.

>+ nsCString proxyMode;
>+ nsCOMPtr<nsIGSettingsCollection> proxy_settings;
>+ mGSettings->GetCollectionForSchema(NS_LITERAL_CSTRING("org.gnome.system.proxy"),
>+ getter_AddRefs(proxy_settings));
>+ if (proxy_settings) {
>+ nsresult rv = proxy_settings->GetString(NS_LITERAL_CSTRING("mode"), proxyMode);

Move the proxyMode declaration to within the "if (proxy_setttings)" block
where it is used.

>+ return proxy_settings->GetString(NS_LITERAL_CSTRING("autoconfig-url"),
>+ aResult);

Alignment.

>+ PRInt32 port;
>+ rv = proxy_settings->GetInt(NS_LITERAL_CSTRING("port"), &port);
>+ NS_ENSURE_SUCCESS(rv, rv);
>+
>+ SetProxyResult(aType, host, port, aResult);
>+ return NS_OK;

Need to return NS_ERROR_FAILURE when the port is 0.

'Each of the 4 proxy types is enabled if its "host" key is
non-empty and its "port" key is non-0.'

http://git.gnome.org/browse/gsettings-desktop-schemas/tree/schemas/org.gnome.system.proxy.gschema.xml.in.in

>+ // Check if proxy is enabled, flag is only in schema org.gnome.system.proxy.http,
>+ // there is no separate flag for each schema.
>+ nsresult rv = mGSettings->GetCollectionForSchema(NS_LITERAL_CSTRING("org.gnome.system.proxy.http"),
>+ getter_AddRefs(proxy_settings));
>+ NS_ENSURE_SUCCESS(rv, rv);
>+ rv = proxy_settings->GetBoolean(NS_LITERAL_CSTRING("enabled"), &masterProxySwitch);
>+ NS_ENSURE_SUCCESS(rv, rv);

"enabled" is described as "Unused", so don't use this key.

>+ bool useHttpProxyForAll = false;
>+ // This setting sometimes doesn't exist, don't bail on failure
>+ proxy_settings->GetBoolean(NS_LITERAL_CSTRING("use-same-proxy"), &useHttpProxyForAll);
>+
>+ if (!useHttpProxyForAll) {
>+ rv = SetProxyResultFromGSettings("org.gnome.system.proxy.socks", "SOCKS", a...

Comment on attachment 577534
gsettings proxy patch v2

>   FUNC(g_variant_get_string, const char *, (GVariant* value, gsize* length)) \
>   FUNC(g_variant_is_of_type, gboolean, (GVariant* value, const GVariantType* type)) \
>   FUNC(g_variant_new_int32, GVariant *, (gint32 value)) \
>   FUNC(g_variant_new_boolean, GVariant *, (gboolean value)) \
>   FUNC(g_variant_new_string, GVariant *, (const char* string)) \
>+  FUNC(g_variant_get_strv, GVariant *, (gsize* length)) \

These were roughly in an order.
Can you insert get_strv after get_string, please?

(In reply to jhorak from comment #10)
> Could you be more specific (the other things)?

Check the #define statements before and after GSETTIINGS_FUNCTIONS as any new
symbols need to be added here.

A #define is needed for g_variant_get_strv and G_VARIANT_TYPE_STRING_ARRAY.

>+  if (mGConf && IsProxyMode("auto")) {
>+    return mGConf->GetString(NS_LITERAL_CSTRING("/system/proxy/autoconfig_url"),
>+                             aResult);
>   }

>+  if (mGSettings) {
>+     // Check if mode is auto

I assume GSettings should override GConf settings.
Otherwise I assume those who upgrade to GNOME 3 will still be using their old
GConf settings, but the configuration utility will change the GSettings
values.

>+     nsCString proxyMode; 
>+     nsCOMPtr<nsIGSettingsCollection> proxy_settings;
>+     mGSettings->GetCollectionForSchema(NS_LITERAL_CSTRING("org.gnome.system.proxy"), 
>+                                        getter_AddRefs(proxy_settings));
>+     if (proxy_settings) {
>+       nsresult rv = proxy_settings->GetString(NS_LITERAL_CSTRING("mode"), proxyMode);

Move the proxyMode declaration to within the "if (proxy_setttings)" block
where it is used.

>+         return proxy_settings->GetString(NS_LITERAL_CSTRING("autoconfig-url"), 
>+                                             aResult);

Alignment.

>+  PRInt32 port;
>+  rv = proxy_settings->GetInt(NS_LITERAL_CSTRING("port"), &port);
>+  NS_ENSURE_SUCCESS(rv, rv);
>+    
>+  SetProxyResult(aType, host, port, aResult);
>+  return NS_OK;

Need to return NS_ERROR_FAILURE when the port is 0.

'Each of the 4 proxy types is enabled if its "host" key is
        non-empty and its "port" key is non-0.'

http://git.gnome.org/browse/gsettings-desktop-schemas/tree/schemas/org.gnome.system.proxy.gschema.xml.in.in

>+  // Check if proxy is enabled, flag is only in schema org.gnome.system.proxy.http,
>+  // there is no separate flag for each schema.
>+  nsresult rv = mGSettings->GetCollectionForSchema(NS_LITERAL_CSTRING("org.gnome.system.proxy.http"),
>+                                                   getter_AddRefs(proxy_settings));
>+  NS_ENSURE_SUCCESS(rv, rv);
>+  rv = proxy_settings->GetBoolean(NS_LITERAL_CSTRING("enabled"), &masterProxySwitch);
>+  NS_ENSURE_SUCCESS(rv, rv);

"enabled" is described as "Unused", so don't use this key.

>+  bool useHttpProxyForAll = false;
>+  // This setting sometimes doesn't exist, don't bail on failure
>+  proxy_settings->GetBoolean(NS_LITERAL_CSTRING("use-same-proxy"), &useHttpProxyForAll);
>+
>+  if (!useHttpProxyForAll) {
>+    rv = SetProxyResultFromGSettings("org.gnome.system.proxy.socks", "SOCKS", aResult);
>+    if (NS_SUCCEEDED(rv))
>+      return rv;
>+  }
>+

Similarly, "use-same-proxy" is also "Unused".

>+  if (aScheme.LowerCaseEqualsLiteral("http") || useHttpProxyForAll) {
>+    rv = SetProxyResultFromGSettings("org.gnome.system.proxy.http", "PROXY", aResult);
>+  } else if (aScheme.LowerCaseEqualsLiteral("https")) {
>+    rv = SetProxyResultFromGSettings("org.gnome.system.proxy.https", "PROXY", aResult);

Need to handle this case:

"If an http proxy is configured, but an https proxy is not,
        then the http proxy is also used for https."

Revision history for this message

In Mozilla Bugzilla #682832, Karlt (karlt) wrote on 2011-12-01:

#17

Comment on attachment 577534
gsettings proxy patch v2

>+ obj->SetData(nsCString(*p_gs_strings));

nsDependentCString is more conventional here.

>-static bool GConfIgnoreHost(const nsACString& aIgnore,
>- const nsACString& aHost)
>+static PRBool HostIgnoredByProxy(const nsACString& aIgnore,
>+ const nsACString& aHost)

We now use bool/true/false.

>+ nsDependentCSubstring aIgnoreStripped(start, slash);

The a- prefix on variables is used for parameters (arguments).
As this is not a parameter, so call it "ignoreStripped".

(In reply to Karl Tomlinson (:karlt) from comment #11)
> >+ if (mGConf && IsProxyMode("auto")) {
> >+ return mGConf->GetString(NS_LITERAL_CSTRING("/system/proxy/autoconfig_url"),
> >+ aResult);
> > }
>
> >+ if (mGSettings) {
> >+ // Check if mode is auto
>
> I assume GSettings should override GConf settings.
> Otherwise I assume those who upgrade to GNOME 3 will still be using their old
> GConf settings, but the configuration utility will change the GSettings
> values.

The existance of GSettings does not imply that gsettings-desktop-schemas is
installed, but, if there is an org.gnome.system.proxy schema, and mode is not
auto, then best to respect the desktop-schemas settings by not falling back to
GConf here.

>+ if (mGSettings)
>+ return GetProxyFromGSettings(scheme, host, port, aResult);
>+ else
>+ return GetProxyFromGConf(scheme, host, port, aResult);

Also here, best to fall back to GConf if org.gnome.system.proxy does not
exist.

Revision history for this message

In Mozilla Bugzilla #682832, Jhorak (jhorak) wrote on 2011-12-01:

#18

Created attachment 578249
gsettings proxy patch v3

Attaching next version. Mentioned issues should be resolved. Also changed UUID of nsIGSettingsService due to its change.

Revision history for this message

In Mozilla Bugzilla #682832, Karlt (karlt) wrote on 2011-12-02:

#19

Comment on attachment 578249
gsettings proxy patch v3

>+ nsCOMPtr<nsIGSettingsCollection> proxy_settings;
>+
>+ // Check if proxy is enabled, flag is only in schema org.gnome.system.proxy.http,
>+ // there is no separate flag for each schema.
>+ nsresult rv = mGSettings->GetCollectionForSchema(NS_LITERAL_CSTRING("org.gnome.system.proxy.http"),
>+ getter_AddRefs(proxy_settings));
>+ NS_ENSURE_SUCCESS(rv, rv);
>+
>+ rv = mGSettings->GetCollectionForSchema(NS_LITERAL_CSTRING("org.gnome.system.proxy"),
>+ getter_AddRefs(proxy_settings));

The "org.gnome.system.proxy.http" schema is now not needed here.

>- if (!mGConf)
>+ if (!mGConf || !mGSettings)
> return GetProxyFromEnvironment(scheme, host, port, aResult);

This isn't what we want. This will use the environment unless GSettings and
GConf are available. Best to treat this as a fallback after the GSettings and
GConf code.

>+ /* Check for org.gnome.syste.proxy schema */
>+ nsCOMPtr<nsIGSettingsCollection> proxy_settings;
>+ mGSettings->GetCollectionForSchema(NS_LITERAL_CSTRING("org.gnome.system.proxy"),
>+ getter_AddRefs(proxy_settings));
>+ /* If schema is found let GSettings determine the right proxy, otherwise fallback to gconf. */
>+ if (proxy_settings)
>+ return GetProxyFromGSettings(scheme, host, port, aResult);

GetCollectionForSchema is an expensive operation, so I don't want the
"org.gnome.system.proxy" schema fetched both here and in GetProxyFromGSettings.

I think it should be fine to fall back to GConf and/or environment variables
when GetProxyFromGSettings returns a failure code.

Revision history for this message

In Mozilla Bugzilla #682832, Jhorak (jhorak) wrote on 2011-12-02:

#20

Created attachment 578520
gsettings proxy patch v4

- Removed forgotten org.gnome.system.proxy.http
- GetProxyFromEnvironment moved to the end as the last method to use to determine proxy
- Part for checking org.gnome.syste.proxy schema removed.

Revision history for this message

In Mozilla Bugzilla #682832, Karlt (karlt) wrote on 2011-12-05:

#21

Comment on attachment 578520
gsettings proxy patch v4

>+ items->AppendElement(obj, PR_FALSE);

Can you update PR_FALSE with false, please?

Revision history for this message

In Mozilla Bugzilla #682832, Jhorak (jhorak) wrote on 2011-12-05:

#22

Created attachment 579030
gsettings proxy patch v5

Thanks for review, attaching patch with replaced PR_FALSE.

Revision history for this message

In Mozilla Bugzilla #682832, Dao-g (dao-g) wrote on 2011-12-05:

#23

http://hg.mozilla.org/integration/mozilla-inbound/rev/2f33758829d2

Revision history for this message

In Mozilla Bugzilla #682832, Matt Brubeck (mbrubeck) wrote on 2011-12-05:

#24

https://hg.mozilla.org/mozilla-central/rev/2f33758829d2

Bug Watch Updater (bug-watch-updater) on 2011-12-08

Changed in firefox:
status:	In Progress → Fix Released

Revision history for this message

CuteChaps (sh-senthilkumar) wrote on 2011-12-08:

#25

How to apply the fix

Chris Coulson (chrisccoulson) on 2012-01-11

Changed in firefox (Ubuntu):
status:	Triaged → Fix Committed

Revision history for this message

Launchpad Janitor (janitor) wrote on 2012-02-03:

#26

Download full text (4.1 KiB)

This bug was fixed in the package firefox - 11.0~b1+build1-0ubuntu1

---------------
firefox (11.0~b1+build1-0ubuntu1) precise; urgency=low

  * New upstream release from the beta channel (FIREFOX_11_0b1_BUILD1)
    - Fix LP: #875266 - Firefox ignores GNOME3 proxy settings
    - Fix LP: #857153 - Needs to get accessibility settings from GSettings

  * Update globalmenu-extension to 2.0.3
  * Ensure that the crash reporter is disabled if rebuilt by Ubuntu
    derivatives, as there will be no crash symbols for those
    - update debian/rules
  * Only add "Ubuntu" to the UA string when being built for Ubuntu
    - update debian/rules
  * Drop obsolete Debian menu file
    - remove debian/firefox.menu.in
    - don't create a 32x32 xpm icon in debian/rules
    - drop the imagemagick build-depend in debian/control
  * Temporarily disable ipdl tests due to build failures. These aren't
    enabled upstream, anyway
    - update debian/config/mozconfig.in
  * Always set the update channel - not setting it at build-time on release
    builds breaks the extensions.checkCompatibility pref. The only things
    using it at runtime are nsBlocklistService, Test Pilot (beta + aurora)
    and the about dialog (where the channel is hidden anyway)
    - update debian/rules
    - update debian/firefox.install.in
  * Don't declare an extra DEB_ENABLE_THUMB2 variable, as it's only used
    for the mozconfig. Just do the "if DEB_HOST_ARCH == armel" check
    directly there instead
    - update debian/rules
    - update debian/config/mozconfig.in
  * Fix LP: #898883 - IPC xpcshell tests hang the buildd's. Give all
    xpcshell tests an X display, as plugin-container won't work without one
    - update debian/build/testsuite.mk
  * Turn on all IPC xpcshell tests again
    - update debian/build/testsute.mk
  * Drop the default-apps xml file from lucid and maverick - there is
    already one provided by gnome-control-center, which means that ours
    is only relevant for nightly builds. It's not worth the extra
    complexity for this
    - remove debian/firefox.xml.in
    - update debian/firefox-gnome-support.install.in
    - update debian/rules
  * Ship Test Pilot as a distribution addon, like upstream. This means
    that the addon manager can update it. It does also mean that it will
    remain installed in users profiles if they try the beta or aurora
    builds, but the Feedback button is disabled on release builds
    - update debian/firefox.install.in
    - fixes LP: #913357
  * Drop patches fixed upstream
    - remove debian/patches/fix-build-failure-without-yarr-jit.patch
    - remove debian/patches/fix-cursor-handling.patch
    - update debian/patches/series
  * Keep the firefox-kde-support suggest on releases older than precise
    for now
    - update debian/rules
  * Ensure that we suggest kmozillahelper on lucid
    - update debian/rules
  * Ensure that we replace kubuntu-firefox-installer on lucid
    - update debian/rules
  * Don't build with --disable-gconf on precise and newer. There won't be
    a hard runtime requirement on this from Firefox 12 anyway, and this
    keeps us closer to the upstream configuration
    - update debian/config/mozconfig.in
...

Affects		Status	Importance	Assigned to	Milestone
	Mozilla Firefox	Fix Released	Medium	mozilla-bugs #682832
	firefox (Ubuntu)	Fix Released	Medium	Unassigned

Mozilla Firefox

Firefox ignores GNOME 3 proxy settings

Bug Description

Related branches

Duplicates of this bug

Other bug subscribers

Bug attachments

Remote bug watches

Changed in firefox (Ubuntu):
status:	Fix Committed → Fix Released