platform/upstream/libsoup.git
11 years agoSoupSession: break infinite loops
Dan Winship [Wed, 6 Mar 2013 17:58:29 +0000 (12:58 -0500)]
SoupSession: break infinite loops

Every year or two some bug comes up that makes libsoup retry a request
infinitely. (Also, apps can do this on their own by not paying
attention to the "retrying" flag in SoupSession::authenticate.) Move
the "too many redirects" code and rework it to handle all possible
cases of "message gets resent a suspicious number of times".

11 years ago2.41.91 2.41.91 LIBSOUP_2_41_91 upstream/2.41.91
Dan Winship [Tue, 5 Mar 2013 15:48:24 +0000 (10:48 -0500)]
2.41.91

11 years agosoup-cache: update cached headers on revalidations
Sergio Villar Senin [Mon, 4 Mar 2013 16:58:41 +0000 (17:58 +0100)]
soup-cache: update cached headers on revalidations

SoupCache was not updating the cached headers on conditional requests. We
were only doing it for revalidations started by libsoup clients.

This also properly reset the values of freshness_lifetime and
must_revalidate on revalidations. These two fields were keeping their
original values even if the server wasn't providing such information.

Finally this adds a new cache test (do_header_test) and fixes the one
disabled with #ifdefs (second revalidations).

https://bugzilla.gnome.org/show_bug.cgi?id=695121

11 years agobuild: Remove gio resource compiler definition from configure.ac
Carlos Garcia Campos [Fri, 1 Mar 2013 13:18:39 +0000 (14:18 +0100)]
build: Remove gio resource compiler definition from configure.ac

The variable GLIB_COMPILE_RESOURCES is already set by AM_PATH_GLIB_2_0
macro. Spotted by Csaba Osztrogonác.

https://bugzilla.gnome.org/show_bug.cgi?id=694942

11 years agosoup-message-queue: hold a reference to the async context
Sergio Villar Senin [Mon, 4 Mar 2013 10:26:17 +0000 (11:26 +0100)]
soup-message-queue: hold a reference to the async context

This will ensure that the GMainContext is not freed at least until we get
rid of the queue item. This was causing crashes when synchronously
retrieving resources in WebKit as it uses a different GMainContext for each
of those synchronous requests.

https://bugzilla.gnome.org/show_bug.cgi?id=694920

11 years ago[l10n] Updated German translation
Mario Blättermann [Sun, 3 Mar 2013 20:11:04 +0000 (21:11 +0100)]
[l10n] Updated German translation

11 years agoUpdated Vietnamese translation
Nguyễn Thái Ngọc Duy [Sat, 2 Mar 2013 13:44:41 +0000 (20:44 +0700)]
Updated Vietnamese translation

11 years agopo/vi: import from Damned Lies
Nguyễn Thái Ngọc Duy [Sat, 2 Mar 2013 13:42:54 +0000 (20:42 +0700)]
po/vi: import from Damned Lies

11 years agoUpdated Traditional Chinese translation(Hong Kong and Taiwan)
Chao-Hsiung Liao [Fri, 1 Mar 2013 14:41:18 +0000 (22:41 +0800)]
Updated Traditional Chinese translation(Hong Kong and Taiwan)

11 years agoPunjabi: Translation updated (aalam)
A S Alam [Tue, 26 Feb 2013 02:07:04 +0000 (07:37 +0530)]
Punjabi: Translation updated (aalam)

11 years agoUpdated slovak translation
Pavol Klačanský [Mon, 25 Feb 2013 19:30:40 +0000 (20:30 +0100)]
Updated slovak translation

11 years agoUpdated Norwegian bokmål translation
Kjartan Maraas [Mon, 25 Feb 2013 16:07:20 +0000 (17:07 +0100)]
Updated Norwegian bokmål translation

11 years agoUpdated Lithuanian translation
Aurimas Černius [Sun, 24 Feb 2013 19:55:55 +0000 (21:55 +0200)]
Updated Lithuanian translation

11 years agoexamples: use GOptionContext (finally)
Dan Winship [Sat, 23 Feb 2013 01:43:33 +0000 (02:43 +0100)]
examples: use GOptionContext (finally)

11 years agoexamples: move examples from tests/ to examples/
Dan Winship [Sat, 23 Feb 2013 01:13:26 +0000 (02:13 +0100)]
examples: move examples from tests/ to examples/

get, simple-httpd, and simple-proxy are more example code than test
programs, so move them into a separate directory.

Also, remove "dns", which was once a sort-of test of SoupAddress, but
is now just a redundant sort-of test of GResolver.

11 years agobuild: remove some cruft
Dan Winship [Fri, 22 Feb 2013 21:13:33 +0000 (22:13 +0100)]
build: remove some cruft

Remove accumulated cruft, mostly from things we used to depend on but
don't any more.

Also a few minor drive-by configure cleanups

11 years agoUpdated Uyghur translation
Gheyret Kenji [Fri, 22 Feb 2013 13:20:48 +0000 (22:20 +0900)]
Updated Uyghur translation

Signed-off-by: Gheyret Kenji <gheyret@gmail.com>
11 years agoUpdated Hebrew translation.
Yaron Shahrabani [Thu, 21 Feb 2013 17:55:39 +0000 (19:55 +0200)]
Updated Hebrew translation.

11 years agoUpdated Galician translations
Fran Diéguez [Wed, 20 Feb 2013 09:40:04 +0000 (10:40 +0100)]
Updated Galician translations

11 years agotests: do skipped tests properly
Dan Winship [Wed, 20 Feb 2013 00:21:35 +0000 (19:21 -0500)]
tests: do skipped tests properly

automake interprets exit code 77 as meaning "skipped", so have the
various apache-dependent, php-dependent, and curl-dependent tests do
that, and compile them unconditionally. (Although, to avoid "unused"
warnings, we end up #ifdeffing out the whole file.)

11 years agotests: [CENSORED] [CENSORED] automake [CENSORED]
Dan Winship [Wed, 20 Feb 2013 00:02:59 +0000 (19:02 -0500)]
tests: [CENSORED] [CENSORED] automake [CENSORED]

Specifying serial-tests in AM_INIT_AUTOMAKE breaks the build with
automake 1.11, so the only way to support both 1.11 and 1.13 is to
make the tests work under the parallel harness. Fortunately this
wasn't that hard.

11 years agoAdded Aragonese translation
Daniel Martinez [Tue, 19 Feb 2013 22:17:42 +0000 (23:17 +0100)]
Added Aragonese translation

11 years agotests: fix under automake 1.13
Dan Winship [Tue, 19 Feb 2013 15:34:22 +0000 (10:34 -0500)]
tests: fix under automake 1.13

The parallel test harness (which is the default as of automake 1.13)
doesn't let you use make functions in the definition of TESTS. (It
generates an invalid Makefile in this case.) Since the tests as
currently written won't work with the parallel harness anyway (since
each apache-based test stops apache when it's done), just force the
serial harness for now.

https://bugzilla.gnome.org/show_bug.cgi?id=694135

11 years agosoup-session.c: do not emit content-sniffed if it isn't cached
Sergio Villar Senin [Tue, 19 Feb 2013 14:44:17 +0000 (15:44 +0100)]
soup-session.c: do not emit content-sniffed if it isn't cached

The SoupCache does not get any information from sniffed content types as it
works at a lower level. That's why sometimes we might not have any
Content-Type information for a cached resource (if the server does not
specify it).

11 years agoUpdated Spanish translation
Daniel Mustieles [Tue, 19 Feb 2013 11:43:10 +0000 (12:43 +0100)]
Updated Spanish translation

11 years agoUpdated Czech translation
Marek Černocký [Tue, 19 Feb 2013 10:23:40 +0000 (11:23 +0100)]
Updated Czech translation

11 years ago2.41.90 2.41.90 LIBSOUP_2_41_90 upstream/2.41.90
Dan Winship [Mon, 18 Feb 2013 23:25:25 +0000 (18:25 -0500)]
2.41.90

11 years agoUpdated Serbian translation
Мирослав Николић [Mon, 18 Feb 2013 19:14:22 +0000 (20:14 +0100)]
Updated Serbian translation

11 years agoUpdated Polish translation
Piotr Drąg [Mon, 18 Feb 2013 15:17:37 +0000 (16:17 +0100)]
Updated Polish translation

11 years agocache-test: added a new cancellation test
Sergio Villar Senin [Fri, 15 Feb 2013 20:40:37 +0000 (21:40 +0100)]
cache-test: added a new cancellation test

Added do_refcounting_test that checks that streams do not leak any
reference when a message is cancelled just before starting to read.

This test also need a new cancel flag called
SOUP_TEST_REQUEST_CANCEL_AFTER_SEND_FINISH.

https://bugzilla.gnome.org/show_bug.cgi?id=682527

11 years agotest-utils: Added SOUP_TEST_REQUEST_CANCEL_IMMEDIATE flag
Sergio Villar Senin [Fri, 15 Feb 2013 20:15:08 +0000 (21:15 +0100)]
test-utils: Added SOUP_TEST_REQUEST_CANCEL_IMMEDIATE flag

We were assuming that if the SOUP_TEST_REQUEST_CANCEL_SOON flag was not
present then we wanted an immediate cancelation. That worked but now we need
to add it to support more cancellation types.

https://bugzilla.gnome.org/show_bug.cgi?id=682527

11 years agosoup-cache-input-stream: fixed a reference cycle
Sergio Villar Senin [Fri, 15 Feb 2013 20:40:28 +0000 (21:40 +0100)]
soup-cache-input-stream: fixed a reference cycle

Fixed a reference cycle between SoupCacheInputStream and a
GTask. This reference cycle was preventing the input stream from
being freed when the SoupMessage was cancelled before the stream
was completely read. Furthermore the SoupCache was not able to
complete the pending request because the SoupCacheInputStream was
not aware of the message cancellation and remained waiting forever.

In order to remove the cycle
soup_cache_input_stream_cache(_finish) were removed and the
caching operation finalization is now reported via a signal.
SoupCacheInputStream also implements close_fn which properly
finalizes any pending caching operation.

https://bugzilla.gnome.org/show_bug.cgi?id=682527

11 years agotests: simplify and improve "make check"
Dan Winship [Sat, 16 Feb 2013 23:26:54 +0000 (18:26 -0500)]
tests: simplify and improve "make check"

Remove rules that automake can figure out itself. Run the tests in
alphabetical order. Run "make kill-httpd" at the end of "make check",
in case the last httpd-using test crashed.

11 years agoSoupSession: remove some dead code, fix proxy error messages
Dan Winship [Sun, 27 Jan 2013 01:06:50 +0000 (20:06 -0500)]
SoupSession: remove some dead code, fix proxy error messages

SoupMessageQueueItem's proxy_addr and proxy_uri fields were no longer
being used, but they were still there, and SoupSession was trying to
use item->proxy_uri to provide clarifying information on
SOUP_STATUS_CANT_RESOLVE_PROXY / SOUP_STATUS_CANT_CONNECT_PROXY.

Remove the unused fields, and fix SoupSession to get the proxy name
from the SoupConnection instead.

https://bugzilla.gnome.org/show_bug.cgi?id=680273

11 years agoInternal renamings for clarity
Dan Winship [Sun, 27 Jan 2013 00:44:56 +0000 (19:44 -0500)]
Internal renamings for clarity

The GSocketClient::event signal gets proxied from GSocketClient to
SoupSocket to SoupConnection to SoupSession to SoupMessage (!) via
signal handlers that used to have "proxy" in their names. But that
made it sound like they had something to do with network proxies. So
rename them all (to "re_emit").

https://bugzilla.gnome.org/show_bug.cgi?id=680273

11 years agoUpdated Malayalam Localization
Anish A [Fri, 15 Feb 2013 06:30:05 +0000 (10:30 +0400)]
Updated Malayalam Localization

11 years agocache-test: added cancellation tests
Sergio Villar Senin [Wed, 30 Jan 2013 17:16:32 +0000 (18:16 +0100)]
cache-test: added cancellation tests

Added cancellation tests for both fresh and under revalidation resources.
These tests ensure that cancellations happen flawlessly and that any
potential revalidation process is also cancelled.

https://bugzilla.gnome.org/show_bug.cgi?id=692310

11 years agotest-utils: add cancellation support to soup_test_request_send
Sergio Villar Senin [Wed, 30 Jan 2013 16:56:38 +0000 (17:56 +0100)]
test-utils: add cancellation support to soup_test_request_send

The function gets a new parametter used to enable request cancellation for
both sync and async sessions. The cancellation could be performed by either
using the GCancellable or by directly cancelling the SoupMessage.

Also the GMainContext used to simulate sync operations with async ones will
now try to execute all its pending events before quiting the main loop.

https://bugzilla.gnome.org/show_bug.cgi?id=692310

11 years agosoup-session: handle cancellation of SoupCache resources
Sergio Villar Senin [Wed, 30 Jan 2013 19:05:29 +0000 (20:05 +0100)]
soup-session: handle cancellation of SoupCache resources

SoupSession will now properly handle cancellations (either with
soup_session_cancel_message or g_cancellable_cancel) of resources returned
by the cache whether they're fresh or need a revalidation.

http://bugzilla.gnome.org/show_bug.cgi?id=692310

11 years agoUpdated slovak translation
Pavol Klačanský [Sat, 9 Feb 2013 21:41:00 +0000 (22:41 +0100)]
Updated slovak translation

11 years agoSoupAuthNTLM: fix to do a "retrying" authenticate
Dan Winship [Thu, 7 Feb 2013 20:03:41 +0000 (15:03 -0500)]
SoupAuthNTLM: fix to do a "retrying" authenticate

If the first attempt at NTLM auth fails, let the auth manager emit a
"retrying" authenticate as well, just like normal auths do.

https://bugzilla.gnome.org/show_bug.cgi?id=693222

11 years agoSoupSession: fix a bug with the add-feature-by-type property
Dan Winship [Thu, 7 Feb 2013 19:14:47 +0000 (14:14 -0500)]
SoupSession: fix a bug with the add-feature-by-type property

The add-feature-by-type and remove-feature-by-type properties were
only accepting SoupSessionFeature types, so you couldn't use it for
things like request types, auth types, etc.

11 years agotests: fix non-php apache tests to still work without php
Dan Winship [Thu, 7 Feb 2013 14:23:31 +0000 (09:23 -0500)]
tests: fix non-php apache tests to still work without php

Setting "IF_HAVE_PHP" to "#" actually sets it to "", because the "#"
is treated as the start of a comment. We have to set it to "\#" if we
want to cause the "#" to actually get substituted into httpd.conf

https://bugzilla.gnome.org/show_bug.cgi?id=693311

11 years agoSoupSession: added a new construction-time property "local-address"
Jonh Wendell [Tue, 5 Feb 2013 15:57:44 +0000 (13:57 -0200)]
SoupSession: added a new construction-time property "local-address"

It will be proxied to the SoupConnection and then to SoupSocket,
allowing the user to specify a local address to bind the local socket.

11 years agoSoupConnection: added a new construction-time property "local-address"
Jonh Wendell [Tue, 5 Feb 2013 15:38:09 +0000 (13:38 -0200)]
SoupConnection: added a new construction-time property "local-address"

It will be proxied to the SoupSocket upon its creation.

11 years agoSoupSocket: Use the local socket, when available, even on client side.
Jonh Wendell [Tue, 5 Feb 2013 00:21:12 +0000 (22:21 -0200)]
SoupSocket: Use the local socket, when available, even on client side.

This allows binding the client connection to a specific local address.

11 years agoSoupSocket: Move the creation of socket clients to a single place
Jonh Wendell [Mon, 4 Feb 2013 23:52:26 +0000 (21:52 -0200)]
SoupSocket: Move the creation of socket clients to a single place

Thus avoiding duplication of code in sync and async functions.

11 years ago2.41.5 2.41.5 LIBSOUP_2_41_5 upstream/2.41.5
Dan Winship [Tue, 5 Feb 2013 15:14:44 +0000 (10:14 -0500)]
2.41.5

11 years agoUpdated Traditional Chinese translation(Hong Kong and Taiwan)
Chao-Hsiung Liao [Mon, 4 Feb 2013 12:34:27 +0000 (20:34 +0800)]
Updated Traditional Chinese translation(Hong Kong and Taiwan)

11 years agosoup-session: use SoupMessageQueueItem's cancellable for the GTask
Sergio Villar Senin [Wed, 30 Jan 2013 18:25:12 +0000 (19:25 +0100)]
soup-session: use SoupMessageQueueItem's cancellable for the GTask

Even if the caller does not specify a GCancellable the GTask must use the
one that comes with the item by default.

This also sets check_cancellable() to FALSE in the session's GTask in order
not to override error messages previously set on the GTask in case of
cancellations.

https://bugzilla.gnome.org/show_bug.cgi?id=692310

11 years ago[l10n] Updated German translation
Andre Jonas [Sun, 3 Feb 2013 18:58:29 +0000 (19:58 +0100)]
[l10n] Updated German translation

11 years ago[l10n] Updated Italian translation.
Milo Casagrande [Thu, 24 Jan 2013 20:48:32 +0000 (21:48 +0100)]
[l10n] Updated Italian translation.

11 years agoRevert "SoupURI: replace NULL password with "" on http"
Dan Winship [Mon, 21 Jan 2013 15:50:16 +0000 (10:50 -0500)]
Revert "SoupURI: replace NULL password with "" on http"

This broke CalDAV in evolution. I'm not sure exactly why, but if it
broke that, it might break other things too, so let's just revert
this.

This reverts commit 38901ca5e684a8fac75e6ff740d45c82dd44181a.

https://bugzilla.gnome.org/show_bug.cgi?id=692149

11 years agoUpdated Norwegian bokmål translation
Kjartan Maraas [Mon, 21 Jan 2013 11:32:31 +0000 (12:32 +0100)]
Updated Norwegian bokmål translation

11 years agoUpdated Uyghur translation
Gheyret Kenji [Sun, 20 Jan 2013 00:27:21 +0000 (09:27 +0900)]
Updated Uyghur translation

Signed-off-by: Gheyret Kenji <gheyret@gmail.com>
11 years agosoup-body-input-stream: update priv->eof when skipping
Sergio Villar Senin [Fri, 18 Jan 2013 18:06:48 +0000 (19:06 +0100)]
soup-body-input-stream: update priv->eof when skipping

We must update the priv->eof field in the skip() operation as we do in
read(). This bug was causing redirects not to finish because the
SoupBodyInputStream was creating a pollable source that will never issue
anything instead of a plain timeout source.

https://bugzilla.gnome.org/show_bug.cgi?id=692026

11 years agoset_ssl_ca_file: don't try unreffing tlsdb if it's NULL
Guillaume Desmottes [Thu, 17 Jan 2013 09:59:13 +0000 (10:59 +0100)]
set_ssl_ca_file: don't try unreffing tlsdb if it's NULL

If g_tls_file_database_new() failed for any reason but
G_TLS_ERROR_UNAVAILABLE, we don't create a fallback DB and so tlsdb stays
NULL.

https://bugzilla.gnome.org/show_bug.cgi?id=691930

11 years agoUpdated Serbian translation
Мирослав Николић [Thu, 17 Jan 2013 08:48:28 +0000 (09:48 +0100)]
Updated Serbian translation

11 years agoUpdated Belarusian translation.
Ihar Hrachyshka [Tue, 15 Jan 2013 16:01:22 +0000 (19:01 +0300)]
Updated Belarusian translation.

11 years ago2.41.4 2.41.4 LIBSOUP_2_41_4
Dan Winship [Mon, 14 Jan 2013 22:25:14 +0000 (17:25 -0500)]
2.41.4

11 years agosoup-session: add SoupMessage-based streaming APIs
Dan Winship [Sat, 29 Dec 2012 02:53:02 +0000 (21:53 -0500)]
soup-session: add SoupMessage-based streaming APIs

SoupRequestHTTP is now implemented in terms of SoupSession SoupMessage
operations, so we might as well just expose those directly, for people
who want streaming APIs, but not SoupRequest.

11 years agosoup-message: belated #include cleanup
Dan Winship [Sat, 12 Jan 2013 14:55:26 +0000 (09:55 -0500)]
soup-message: belated #include cleanup

Remove some includes that I had commented out to see if they were
still needed, but then forgot to actually remove when it turned out
they weren't.

11 years agoSoupURI: replace NULL password with "" on http
Dan Winship [Sun, 9 Dec 2012 12:23:25 +0000 (13:23 +0100)]
SoupURI: replace NULL password with "" on http

All supported HTTP auth methods use both username and password, so
assume a password of "" if none is given

11 years agosoup-auth-digest: fix a leak on re-auth
Dan Winship [Sat, 12 Jan 2013 14:42:50 +0000 (09:42 -0500)]
soup-auth-digest: fix a leak on re-auth

Calling soup_auth_authenticate() on an already-authenticated
SoupAuthDigest would leak the old username and hashed password.

https://bugzilla.gnome.org/show_bug.cgi?id=690142

11 years agoUpdated Bulgarian translation
Alexander Shopov [Sat, 12 Jan 2013 04:56:29 +0000 (06:56 +0200)]
Updated Bulgarian translation

11 years agosoup-session: fix a deadlock when aborting a SoupSessionSync
Dan Winship [Wed, 9 Jan 2013 22:21:13 +0000 (17:21 -0500)]
soup-session: fix a deadlock when aborting a SoupSessionSync

We need to signal conn_cond when removing an item from the queue,
since flush_queue() expects that... (and also, since in some cases,
removing a message from the queue will allow another one to start).

https://bugzilla.gnome.org/show_bug.cgi?id=691399

11 years agoAdd explicit soup-version.h include into soup.h
Milan Crha [Wed, 9 Jan 2013 18:20:01 +0000 (19:20 +0100)]
Add explicit soup-version.h include into soup.h

11 years agobuild: Do not use AM_GLIB_GNU_GETTEXT
Javier [Tue, 8 Jan 2013 17:48:41 +0000 (17:48 +0000)]
build: Do not use AM_GLIB_GNU_GETTEXT

The po/Makefile.in.inis generated by intltoolize

11 years agoautogen.sh: Use autoreconf instead gnome-autogen
Javier [Tue, 8 Jan 2013 16:40:48 +0000 (16:40 +0000)]
autogen.sh: Use autoreconf instead gnome-autogen

11 years agosoup_message_get_https_status: provide info for failed attempts too
Dan Winship [Mon, 7 Jan 2013 21:30:55 +0000 (16:30 -0500)]
soup_message_get_https_status: provide info for failed attempts too

SoupMessage:tls-certificate and SoupMessage:tls-errors were only
getting set for successful https connections. It is useful to have
them be set on failed ones as well. Fix that, and make ssl-test test
it.

https://bugzilla.gnome.org/show_bug.cgi?id=690176

11 years agoRevert "SoupConnection: disconnect on failed connections/tunnels"
Dan Winship [Mon, 7 Jan 2013 20:50:36 +0000 (15:50 -0500)]
Revert "SoupConnection: disconnect on failed connections/tunnels"

Disconnecting clears priv->sock, and in some cases, the caller may
want to look at that after the failed connection. So switch this back
to having the session do the disconnecting itself (which is simpler
now due to various SoupSession rewrites), and remove one inconsistent
case where the connection was disconnecting itself before.

This reverts commit 21d707d66460233596a04f6dc99c945d2dbc3085.

11 years agoFix previous commit
Dan Winship [Sun, 6 Jan 2013 21:38:30 +0000 (16:38 -0500)]
Fix previous commit

11 years agosoup-session: fix up the SoupRequest requeuing logic
Dan Winship [Sat, 5 Jan 2013 23:25:12 +0000 (18:25 -0500)]
soup-session: fix up the SoupRequest requeuing logic

Split the "this message is likely to be requeued" logic into a single
function that the sync and async codepaths can share, and fix it to
not think that SOUP_MESSAGE_NO_REDIRECT messages will be redirected,
or that messages that disable SOUP_TYPE_AUTH_MANAGER will be
authenticated.

11 years agoRename LIBSOUP_DISABLE_DEPRECATED to SOUP_DISABLE_DEPRECATED
Dan Winship [Fri, 4 Jan 2013 15:36:21 +0000 (10:36 -0500)]
Rename LIBSOUP_DISABLE_DEPRECATED to SOUP_DISABLE_DEPRECATED

GNOME_MAINTAINER_MODE_DEFINES apparently sets
LIBSOUP_DISABLE_DEPRECATED, which is now causing various jhbuild
modules to break. "Fix" this by renaming the define (which is only
still there because gtk-doc complains if it's not). The *right* way to
get deprecation warnings/errors now is to define
SOUP_VERSION_MIN_REQUIRED.

11 years agosoup-auth-manager: make this thread-safe
Dan Winship [Tue, 1 Jan 2013 19:18:46 +0000 (14:18 -0500)]
soup-auth-manager: make this thread-safe

Very belatedly, make SoupAuthManager thread-safe.

It also ensures that any SoupAuth that it manages is only manipulated
from one thread at a time.

11 years agosoup-auth-manager: add soup_auth_manager_use_auth()
Dan Winship [Tue, 1 Jan 2013 16:46:18 +0000 (11:46 -0500)]
soup-auth-manager: add soup_auth_manager_use_auth()

Add soup_auth_manager_use_auth(), for "preloading" authentication, and
make the old automatically-request-NTLM behavior happen only for
the legacy SoupSession subclasses.

11 years agoSoupAuthManager: make this public
Dan Winship [Tue, 1 Jan 2013 16:43:28 +0000 (11:43 -0500)]
SoupAuthManager: make this public

11 years agosoup-auth-manager: split out connection handling
Dan Winship [Wed, 12 Dec 2012 09:19:00 +0000 (10:19 +0100)]
soup-auth-manager: split out connection handling

Add a new SoupConnectionAuth class to help with connection tracking,
and make SoupAuthNTLM a subclass of it. Allow a single SoupAuthNTLM to
carry state information about multiple connections.

Make SoupSession store the SoupConnection a SoupMessage is associated
with on the message, and use that from SoupConnectionAuth rather than
tracking sockets by hand like SoupAuthManager had previously done.
Remove the connection tracking in SoupAuthManager, since it is no
longer needed.

11 years agosoup-auth-ntlm: get rid of separate SSO states
Dan Winship [Sun, 30 Dec 2012 15:42:29 +0000 (10:42 -0500)]
soup-auth-ntlm: get rid of separate SSO states

Use a single boolean to track whether we're using single-sign-on
(/usr/bin/ntlm_auth) rather than having two parallel sets of
SoupNTLMStates.

11 years agontlm-test: kill some dead code
Dan Winship [Wed, 2 Jan 2013 02:30:05 +0000 (21:30 -0500)]
ntlm-test: kill some dead code

11 years agontlm-test-helper: fix
Dan Winship [Sun, 30 Dec 2012 23:12:59 +0000 (18:12 -0500)]
ntlm-test-helper: fix

We need to set SOUP_NTLM_AUTH_DEBUG to an empty string (to tell
SoupAuthNTLM to always use built-in NTLM), not leave it unset, which
will let SoupAuthNTLM use the real ntlm_auth binary.

(The effect of this bug was that instead of always using built-in
NTLM, ntlm-test-helper would first try to use ntlm_auth, and fail
because there were no cached credentials, and then fall back to
built-in NTLM, which meant that we were testing the fallback codepath
even in the supposed-to-be-non-fallback case. Unless your username was
"alice" or "bob" and you had a working winbind configuration, in which
case the tests would just fail.)

Also, fix up ntlm-test a bit, so that if ntlm-test-helper for some
reason tries to authenticate as someone other than alice or bob, it
recognizes this, rather than considering everyone who isn't alice to
be bob.

11 years agosoup-auth-domain-digest: minor documentation correction
Guenther Wutz [Mon, 31 Dec 2012 19:28:37 +0000 (20:28 +0100)]
soup-auth-domain-digest: minor documentation correction

11 years agoUpdated Friulian translation
TmTFx [Sat, 29 Dec 2012 01:58:21 +0000 (23:58 -0200)]
Updated Friulian translation

11 years agosoup-auth-ntlm: small fix
Dan Winship [Sun, 30 Dec 2012 00:33:21 +0000 (19:33 -0500)]
soup-auth-ntlm: small fix

11 years agoReorganize NTLM authentication, fix ntlm_auth support
Dan Winship [Tue, 11 Dec 2012 10:10:45 +0000 (11:10 +0100)]
Reorganize NTLM authentication, fix ntlm_auth support

Move the auth-managery parts of SoupAuthManagerNTLM down into
SoupAuthManager, and the NTLMy parts into SoupAuthNTLM (in preparation
for supporting other kinds of connection-based auth such as
Negotiate).

The reorganization also makes it possible to use SoupAuthNTLM to
implement a mock version of /usr/bin/ntlm_auth, so we can extend
ntlm-test to test both the built-in-NTLM and external-NTLM codepaths.
Doing this reveals that, AFAICT, the external codepath did not
previously actually work, because it mis-used
G_SPAWN_FILE_AND_ARGV_ZERO and so ended up passing incorrect arguments
to /usr/bin/ntlm_auth.

11 years agotests: add session-test
Dan Winship [Fri, 14 Dec 2012 20:50:59 +0000 (15:50 -0500)]
tests: add session-test

Add a test that each of the 3 SoupSession types (plain, async, sync)
behaves as expected with soup_session_queue_message(),
soup_session_send_message(), and soup_session_cancel_message().

11 years agoSoupSession: update docs (and some semantics) for plain SoupSession
Dan Winship [Fri, 14 Dec 2012 14:54:36 +0000 (09:54 -0500)]
SoupSession: update docs (and some semantics) for plain SoupSession

Update the docs to reflect the changes in plain SoupSessions vs its
traditional subclasses. And while we're there, add a few new ones
(including making soup_session_queue_message() and
soup_session_send_message() work on plain SoupSession.

11 years agoRevert the mirroring of SoupMessage API onto SoupRequestHTTP.
Dan Winship [Fri, 28 Dec 2012 15:41:20 +0000 (10:41 -0500)]
Revert the mirroring of SoupMessage API onto SoupRequestHTTP.

SoupMessage isn't being deprecated, and mirroring its API onto
SoupRequestHTTP is just going to result in always having to add every
new API twice. Also, it turns out to be less useful than originally
expected anyway, since you end up having to cast between SoupRequest
and SoupRequestHTTP frequently anyway.

This reverts commit d7117329400e47d2187ed033099d921d555f8d71 and most of
commit 53c270d0e2868fa5ad48ce864f10a9486b11a071.

11 years agoAdd back LIBSOUP_DISABLE_DEPRECATED
Dan Winship [Fri, 28 Dec 2012 15:32:11 +0000 (10:32 -0500)]
Add back LIBSOUP_DISABLE_DEPRECATED

gtk-doc expects "Deprecated:" notes in docs to be matched up with some
deprecation-guard #define. So even though we don't want people to use
it any more, add LIBSOUP_DISABLE_DEPRECATED back.

11 years agodocs: more fixes
Dan Winship [Fri, 28 Dec 2012 15:32:30 +0000 (10:32 -0500)]
docs: more fixes

11 years agodocs: Lots of minor fixes and additions
Dan Winship [Fri, 28 Dec 2012 02:13:34 +0000 (21:13 -0500)]
docs: Lots of minor fixes and additions

11 years agoUpdated Hebrew translation.
Yaron Shahrabani [Wed, 26 Dec 2012 22:50:21 +0000 (00:50 +0200)]
Updated Hebrew translation.

11 years agosoup-session: move cache handling here
Dan Winship [Wed, 26 Dec 2012 19:50:11 +0000 (14:50 -0500)]
soup-session: move cache handling here

Move the SoupCache special handling from SoupRequestHTTP to
SoupSession (as another step towards making SoupCache not such a
magical special case hack that only works in certain situations).

11 years agotests: add cache-test
Dan Winship [Tue, 25 Dec 2012 23:17:05 +0000 (18:17 -0500)]
tests: add cache-test

Add a test of basic cache functionality (finally!)

Still needs more tests...

11 years agotest-utils: remove unnecessary #includes
Dan Winship [Tue, 25 Dec 2012 17:04:46 +0000 (12:04 -0500)]
test-utils: remove unnecessary #includes

11 years agoUpdated Polish translation
Piotr Drąg [Sun, 23 Dec 2012 03:29:54 +0000 (04:29 +0100)]
Updated Polish translation

11 years agoUpdated Galician translations
Fran Diéguez [Sun, 23 Dec 2012 03:04:03 +0000 (04:04 +0100)]
Updated Galician translations

11 years agoUpdated Slovenian translation
Matej Urbančič [Sat, 22 Dec 2012 14:22:49 +0000 (15:22 +0100)]
Updated Slovenian translation

11 years agoUpdated Spanish translation
Daniel Mustieles [Thu, 20 Dec 2012 10:10:21 +0000 (11:10 +0100)]
Updated Spanish translation

11 years ago[l10n] Updated Estonian translation
Mattias Põldaru [Wed, 19 Dec 2012 16:22:22 +0000 (18:22 +0200)]
[l10n] Updated Estonian translation