Christian Dywan [Mon, 7 Mar 2011 16:18:40 +0000 (17:18 +0100)]
Parse month names regardless of case
A test case is added that verifies that this works.
Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=644048
Dan Winship [Tue, 22 Feb 2011 02:10:42 +0000 (21:10 -0500)]
2.33.90
Sergio Villar Senin [Wed, 16 Feb 2011 10:27:44 +0000 (11:27 +0100)]
soup-cache.c: do not cache multipart/x-mixed-replace content
That MIME type is normally used to stream video from webcams. Makes no sense to
cache those resources as they are always changing.
https://bugzilla.gnome.org/show_bug.cgi?id=642028
Sergio Villar Senin [Mon, 24 Jan 2011 13:32:11 +0000 (14:32 +0100)]
soup-session.c: send Keep-Alive for HTTP/1.0 connections
Sending the keep alive header prevents connections from closing and thus
improves overall performance when requesting multiple resources from HTTP/1.0
hosts as we do not need to create new connections for each request but reuse
the existing ones.
https://bugzilla.gnome.org/show_bug.cgi?id=640414
Dan Winship [Tue, 15 Feb 2011 20:13:32 +0000 (15:13 -0500)]
soup-message-client-io: ASCIIfy unicode hostnames for Host header
We were resolving unicode hostnames correctly, but then putting them
in the Host header as UTF-8 rather than using the ASCII-encoded
version. Fix that.
https://bugzilla.gnome.org/show_bug.cgi?id=642075
Andre Klapper [Wed, 2 Feb 2011 23:35:25 +0000 (00:35 +0100)]
Fix typo
Dan Winship [Wed, 2 Feb 2011 20:38:26 +0000 (15:38 -0500)]
Make the glib-networking dependency more explicit
Require glib-networking to be present at build time (with an error
explaining that you can pass --disable-tls-check, but packagers should
make sure libsoup has a runtime dependency on glib-networking).
Also, if a TLS connection fails because of missing glib-networking,
mention that in the reason_phrase, as a last resort.
Dan Winship [Mon, 31 Jan 2011 21:47:17 +0000 (16:47 -0500)]
2.33.6
Dan Winship [Tue, 25 Jan 2011 17:23:08 +0000 (12:23 -0500)]
soup-cache: fix a warning (and a bug)
If a request specifies "Cache-Control: max-age=0", revalidate the
response, but don't mark the cache entry as "must_revalidate", since
the need for revalidation applies only to the current request.
https://bugzilla.gnome.org/show_bug.cgi?id=640556
Sergio Villar Senin [Mon, 31 Jan 2011 12:29:22 +0000 (13:29 +0100)]
soup-request-data: return decoded contents for non-base64 data URLs
SoupRequestData was not returning the decoded version of data URLs when they
were not encoded in base64
https://bugzilla.gnome.org/show_bug.cgi?id=641022
Sergio Villar Senin [Fri, 28 Jan 2011 22:32:23 +0000 (17:32 -0500)]
SoupCache fails to load resources when cache contents are externally removed
Try to download cached resources again if they are no longer
accesible by the cache.
https://bugs.webkit.org/show_bug.cgi?id=50577
(belatedly copied from WebKit's version of this file)
Dan Winship [Tue, 25 Jan 2011 17:18:57 +0000 (12:18 -0500)]
Fix a warning (again)
soup-date uses strtoul() to parse the sub-second portion of an ISO
8601 timestamp (to verify that it's syntactically valid), but then
throws that information away. gcc 4.6 is pickier about how we throw it
away though.
Sven Neumann [Tue, 18 Jan 2011 15:22:26 +0000 (10:22 -0500)]
soup-form: fix soup_form_decode() for multiple values with the same key
If multiple values in an urlencoded dataset use the same key, the
resulting value in the hash-table created by soup_form_decode() used
to point to memory that is already freed. That is because
g_hash_table_insert() frees the passed key which happens to point
to the memory that contains the new value. Fix this by using
g_hash_table_replace() instead.
Sven Neumann [Tue, 18 Jan 2011 15:08:41 +0000 (10:08 -0500)]
Add test for soup_form_decode().
Add a test that checks how soup_form_decode() handles multiple values
with the same key.
Sergio Villar Senin [Mon, 17 Jan 2011 17:43:20 +0000 (18:43 +0100)]
soup-session-async:fixed a connection starvation issue
Trying to cleanup connections ASAP could potentially lead to starvation of the
messages in the queue tail if we issue a lot of connections in a short period
of time to a reduced set of servers. As we are able to notice connections
closed by servers without even running the cleanup connections code, trying to
prune connections so quickly is not that important now.
https://bugzilla.gnome.org/show_bug.cgi?id=639768
Dan Winship [Mon, 10 Jan 2011 22:54:33 +0000 (17:54 -0500)]
2.33.5
Dan Winship [Mon, 10 Jan 2011 18:47:52 +0000 (13:47 -0500)]
soup_session_cancel_message: fix up, especially in sync sessions
Cancelling a message from another thread had some race conditions that
could sometimes cause crashes. Fix things up a bit by using
GCancellable to interrupt the I/O, rather than calling
soup_message_io_finished() directly.
Also added a test for this case to tests/misc-test, although
unfortunately due to the raciness of the bug, it only failed
sporadically even before the fix (but seems to fail never now).
https://bugzilla.gnome.org/show_bug.cgi?id=637741
Dan Winship [Mon, 10 Jan 2011 17:50:21 +0000 (12:50 -0500)]
soup-message-io: don't watch for SoupSocket::disconnect
The IO code was explicitly handling the SoupSocket::disconnect signal,
but this is actually redundant; if the socket gets disconnected we'll
get either an error (if writing) or an eof (if reading), and the code
will do the right thing with that. Watching ::disconnected too just
results in processing the same error twice and having to be extra
careful to do it idempotently.
Dan Winship [Tue, 21 Dec 2010 15:39:19 +0000 (10:39 -0500)]
2.33.4
John (J5) Palmieri [Mon, 13 Dec 2010 20:33:16 +0000 (15:33 -0500)]
[gi] annotate nicks for the elements of HTTPVersion enum
* GI was parsing as HTTPVersion.SOUP_HTTP_1_0 as HTTPVersion.0
* GI now parses as HTTPVersion.HTTP_1_0 (identifiers can't start with a number
so we append HTTP for good measure)
Sergio Villar Senin [Mon, 13 Dec 2010 09:31:22 +0000 (10:31 +0100)]
soup-http-input-stream: fixed a condition
priv->caller_buffsize and priv->leftover_bufsize are both unsigned integers,
and thus, checking (a - b > 0) could incorrectly return TRUE if b > a.
Dan Winship [Sun, 12 Dec 2010 11:14:06 +0000 (12:14 +0100)]
soup-request-data: another fix
Dan Winship [Sun, 12 Dec 2010 10:50:57 +0000 (11:50 +0100)]
soup-request-data: Fix content-type decoding
Given "data:text/html,<html>...", it was reporting the content-type as
"data:text/html". Fix.
https://bugs.webkit.org/show_bug.cgi?id=50885
Dan Winship [Sat, 11 Dec 2010 10:31:59 +0000 (11:31 +0100)]
soup-cookie-jar: allow setting cookies on file:// URIs
Because other browsers do, and WebKit has a test for it.
https://bugzilla.gnome.org/show_bug.cgi?id=603825
Jonathon Jongsma [Wed, 8 Dec 2010 05:09:31 +0000 (23:09 -0600)]
Fix leak when processing the set cookie header
Valgrind trace:
==4617== 401 (96 direct, 305 indirect) bytes in 2 blocks are definitely lost in loss record 15,500 of 16,590
==4617== at 0x4C244E8: malloc (vg_replace_malloc.c:236)
==4617== by 0x103D73AA: g_malloc (gmem.c:164)
==4617== by 0x103F0EB4: g_slice_alloc (gslice.c:842)
==4617== by 0x103F0F0C: g_slice_alloc0 (gslice.c:854)
==4617== by 0xE1CCD04: parse_one_cookie (soup-cookie.c:218)
==4617== by 0xE1CD2B3: soup_cookies_from_response (soup-cookie.c:763)
==4617== by 0xE1CE332: process_set_cookie_header (soup-cookie-jar.c:562)
==4617== by 0xFB4AC36: g_cclosure_marshal_VOID__VOID (gmarshal.c:79)
==4617== by 0xFB2F59E: g_closure_invoke (gclosure.c:766)
==4617== by 0xFB4A2B1: signal_emit_unlocked_R (gsignal.c:3252)
==4617== by 0xFB490EE: g_signal_emit_valist (gsignal.c:2983)
==4617== by 0xFB4966C: g_signal_emit (gsignal.c:3040)
==4617== by 0xE1DA0DF: io_read (soup-message-io.c:903)
https://bugzilla.gnome.org/show_bug.cgi?id=636741
Dan Winship [Fri, 10 Dec 2010 11:53:48 +0000 (12:53 +0100)]
Don't uselessly g_warn when TLS isn't available
We g_warn if we can't read the provided SOUP_SESSION_SSL_CA_FILE
(since we have no other way to indicate the error to the app). But
don't warn if the problem is just that the TLS backend isn't available,
since that will cause equally-obvious failures later on that don't
require g_warnings.
Dan Winship [Fri, 10 Dec 2010 11:44:59 +0000 (12:44 +0100)]
Remove SoupCoding, SoupCodingGZip, use GZlibDecompressor
also add a new test to coding-test for handling content that
doesn't decode correctly
Dan Winship [Fri, 10 Dec 2010 11:27:52 +0000 (12:27 +0100)]
fix previous commit; I only tested the "not working" case before...
Dan Winship [Fri, 10 Dec 2010 09:27:39 +0000 (10:27 +0100)]
Add glib-networking to the MISSING_REGRESSION_TEST_PACKAGES checks
and make timeout-test and proxy-test skip their https tests if
it's not available
Sergio Villar Senin [Thu, 9 Dec 2010 16:39:09 +0000 (17:39 +0100)]
SoupDirectoryInputStream: fixed a reference leak
The GFileInfo returned by g_file_enumerator_next_file must be unref'ed when no
longer needed
Sergio Villar Senin [Thu, 9 Dec 2010 12:32:41 +0000 (13:32 +0100)]
Cache must not return a cached resource with "no-cache" directive
If "Pragma: no-cache" or "Cache-Control: no-cache" do exist in a request then
the cache must not return a cached resource but reload the resource from the
original server.
A "Cache-Control: max-age=0" should also force the cache to revalidate its
entries against the original server.
Dan Winship [Thu, 9 Dec 2010 11:08:06 +0000 (12:08 +0100)]
2.33.1 (not a release, just a pkg-config-able number)
2.33.1 is "the version with SoupRequest and SoupCache"
Dan Winship [Wed, 8 Dec 2010 11:52:54 +0000 (12:52 +0100)]
Make SoupRequester a SoupSessionFeature and the request types sub-features
WebKit already exposes the SoupSession to its users, so this change
will automatically give them access to the SoupRequest infrastructure
as well.
Dan Winship [Wed, 8 Dec 2010 09:40:02 +0000 (10:40 +0100)]
Import SoupRequest/cache stuff from new-io branch / WebKit
SoupRequest stuff is based on Gabriel Corvalan's Google Summer of Code
project from 2009, as further hacked on at the 2009 WebKitGTK Hackfest.
The cache stuff was written on top of that by Sergio Villar and
initially committed to WebKit since it only works with the SoupRequest
infrastructure, and the SoupRequest stuff was never fully completed
(eg, the synchronous versions never worked right).
Move all of this here now as a first step to finishing it, but hide it
behind #ifdef LIBSOUP_USE_UNSTABLE_REQUEST_API.
Dan Winship [Mon, 7 Jun 2010 21:03:19 +0000 (17:03 -0400)]
SoupSession, SoupAuthManagerNTLM: Use subfeatures for NTLM auth too
Deprecate SoupSession:use-ntlm in favor of using
soup_session_add_feature_by_type() with SOUP_TYPE_AUTH_NTLM.
Dan Winship [Mon, 7 Jun 2010 20:46:20 +0000 (16:46 -0400)]
SoupAuthManager: implement subfeatures for adding/removing auth types
Allow using soup_session_add/remove_feature_by_type() to add/remove auth
types from the session, and add a test for that
Dan Winship [Mon, 7 Jun 2010 20:45:22 +0000 (16:45 -0400)]
SoupSession: allow features to implement "subfeatures"
Let soup_session_add_feature_by_type() (and remove()) add and remove
"features" from existing features. Eg, to add/remove auth types or
Content-Encodings.
Lucas Rocha [Tue, 7 Dec 2010 15:56:27 +0000 (15:56 +0000)]
soup-message-headers: add missing annotations in get_content_type
Dan Winship [Tue, 7 Dec 2010 14:12:08 +0000 (15:12 +0100)]
Set SoupMessage:tls-certificate from SoupSocket:tls-certificate sooner
Use the new GTlsConnection:peer-certificate semantics to guarantee we
set the certificate on the message as soon as it's been accepted.
Dan Winship [Mon, 6 Dec 2010 13:05:15 +0000 (14:05 +0100)]
Add SoupMessage:tls-certificate and SoupMessage:tls-errors
These provide more information about the certificate on the other end of
a TLS connection.
Dan Winship [Fri, 13 Nov 2009 00:44:27 +0000 (19:44 -0500)]
SoupSocket: port to GSocketConnection/GTlsConnection
and remove libsoup's built-in TLS support, which is no longer needed
Dan Winship [Tue, 7 Dec 2010 10:22:28 +0000 (11:22 +0100)]
Bump version to 2.33.0
Evan Nemerson [Sun, 21 Nov 2010 00:11:39 +0000 (16:11 -0800)]
Include C header and exported packages information in GIRs
https://bugzilla.gnome.org/show_bug.cgi?id=635395
Dan Winship [Mon, 29 Nov 2010 23:01:58 +0000 (18:01 -0500)]
2.32.2
Dan Winship [Tue, 23 Nov 2010 20:50:42 +0000 (15:50 -0500)]
SoupSocket: tweak GMainContext handling
GSocketClient may take multiple steps, in which case we need to make
sure that the socket's async_context is the thread-default context
throughout all of them.
This will break horribly if someone tries to send things off to
multiple different contexts from a single thread, but that was never
supposed to have worked, and always had race conditions anyway.
https://bugzilla.gnome.org/show_bug.cgi?id=635101
Dan Winship [Tue, 23 Nov 2010 17:53:37 +0000 (12:53 -0500)]
SoupSessionSync: fix messages getting stuck forever
Messages in SoupSessionSync were getting stuck forever if they didn't
manage to get a connection on the first try; the test programs didn't
pick this up because none of them queue enough messages all at once to
hit the max-conns limit.
Fix the messages-getting-stuck bug, and add a test case to misc-test
to verify it. Also fix another SoupSessionSync bug where cancelling
several messages at once could cause spurious errors.
https://bugzilla.gnome.org/show_bug.cgi?id=634422
Dan Winship [Thu, 18 Nov 2010 15:15:28 +0000 (10:15 -0500)]
misc-test: fix up the new persistent connection timeout test
The original version of the test implicitly depended on thread
scheduling, and would fail if the server thread didn't get a chance to
run after returning its response but before the the client thread made
its next request. Fix that.
https://bugzilla.gnome.org/show_bug.cgi?id=631525
Dan Winship [Mon, 15 Nov 2010 23:42:02 +0000 (18:42 -0500)]
2.32.1
Dan Winship [Mon, 15 Nov 2010 23:41:43 +0000 (18:41 -0500)]
tests/httpd.conf.in: fix passwords-in-uri stuff for srcdir!=builddir
Sergio Villar Senin [Wed, 6 Oct 2010 16:51:11 +0000 (18:51 +0200)]
Add tests to auth-test for password-in-URI
https://bugzilla.gnome.org/show_bug.cgi?id=631679
Dan Winship [Sun, 14 Nov 2010 16:48:04 +0000 (11:48 -0500)]
soup-auth-manager: make URI username/password override cached info
This allows a page to, eg, force a logout by sending an XMLHttpRequest
using a bad password.
Based on a patch from Sergio Villar.
https://bugzilla.gnome.org/show_bug.cgi?id=631679
Dan Winship [Sat, 13 Nov 2010 22:06:26 +0000 (17:06 -0500)]
Return SOUP_STATUS_CAN_RESOLVE_PROXY as appropriate
We weren't converting the status codes to their proxy versions when
proxy address resolution failed.
https://bugzilla.gnome.org/show_bug.cgi?id=632354
Dan Winship [Wed, 10 Nov 2010 21:48:56 +0000 (16:48 -0500)]
soup-message-io: fix retry-after-unexpected-connection-close
When sending a request on a previously-used connection, we have to
deal with the possibility of the server deciding to time out the
connection right as we start sending data (which sounds like a crazy
race condition, but is in fact pretty much standard behavior). This
got broken in the connection/session reorg earlier in the year. Fix
it.
Also, add a test to misc-test for this.
Based on patch from Sergio Villar.
https://bugzilla.gnome.org/show_bug.cgi?id=631525
Dan Winship [Tue, 9 Nov 2010 18:22:25 +0000 (13:22 -0500)]
soup-message-io: deal with stupid servers that do chunked encoding wrong
Some servers fail to send the final 0-length chunk when allegedly doing
chunked encoding. Browsers cope. Now we do too.
https://bugzilla.gnome.org/show_bug.cgi?id=629160
Dan Winship [Tue, 9 Nov 2010 17:58:31 +0000 (12:58 -0500)]
timeout-test: add SOUP_SESSION_IDLE_TIMEOUT tests as well
Dan Winship [Tue, 9 Nov 2010 17:23:10 +0000 (12:23 -0500)]
Fix parsing/generation of URI queries that aren't HTML forms
soup_form_decode() used to accept strings that weren't "name=value"
pairs, but soup_form_encode_hash() would crash if you passed the
resulting GHashTable to it. Fix both sides: now soup_form_decode()
ignores non-"name=value" elements, and soup_form_encode_hash()
g_return_if_fail()s rather than crashing.
Also fix use of deprecated soup-form.h method names in soup-uri.c and
soup-server.c
https://bugzilla.gnome.org/show_bug.cgi?id=620220
Dan Winship [Fri, 24 Sep 2010 18:07:59 +0000 (14:07 -0400)]
SoupURI: never return NULL from soup_uri_decode/soup_uri_normalize
Normally you just want to DTRT, and soup_uri_new() already behaved this
way anyway.
https://bugzilla.gnome.org/show_bug.cgi?id=630540
Dan Winship [Tue, 9 Nov 2010 14:57:55 +0000 (09:57 -0500)]
SoupContentDecoder: fix an unlikely leak
In particular, this can happen if you accidentally install two content
decoders on the same session. Pointed out a long time ago in email by
Christophe Gillette.
Dan Winship [Thu, 28 Oct 2010 14:42:12 +0000 (10:42 -0400)]
Remove old experimental python bindings
They don't compile any more, and gobject-introspection works now
Dan Winship [Thu, 14 Oct 2010 20:00:05 +0000 (16:00 -0400)]
Appease some gcc 4.5 warnings
Dan Winship [Thu, 14 Oct 2010 19:30:45 +0000 (15:30 -0400)]
SoupCookieJarText: fix a bug when deleting a cookie
based on a patch from Michał Kazior
https://bugzilla.gnome.org/show_bug.cgi?id=631641
Dan Winship [Tue, 28 Sep 2010 00:34:07 +0000 (20:34 -0400)]
2.32.0
Dan Winship [Mon, 13 Sep 2010 19:49:06 +0000 (15:49 -0400)]
2.31.92
Dan Winship [Mon, 13 Sep 2010 14:31:02 +0000 (10:31 -0400)]
Fix a crash when resolving URIs with both spaces and non-UTF8 chars
When using "%.*s" in a UTF-8 locale, in at least some cases, glibc
requires that the string not end in something that looks like a
partial UTF-8 character. This seems wrong according to the c99 spec to
me, but regardless, we need to work around it.
https://bugzilla.gnome.org/show_bug.cgi?id=629449
Dan Winship [Thu, 2 Sep 2010 02:28:56 +0000 (22:28 -0400)]
Fix introspection scanning: don't scan uninstalled .h files
Dan Winship [Thu, 2 Sep 2010 02:02:04 +0000 (22:02 -0400)]
Fixes and additions for new introspection scanner
Dan Winship [Mon, 13 Sep 2010 19:36:31 +0000 (15:36 -0400)]
Remove use of --warn-all with g-ir-scanner
Since it's not possible to fix some of the warnings yet
Colin Walters [Thu, 9 Sep 2010 13:33:07 +0000 (09:33 -0400)]
introspection: Replace usage of --strip-prefix
Update to use --identifier-prefix. Also add --warn-all.
Bump required introspection version.
Sergio Villar Senin [Fri, 3 Sep 2010 19:07:30 +0000 (21:07 +0200)]
Do not uppercase percent-encoded triplets when normalizing URIs
https://bugzilla.gnome.org/show_bug.cgi?id=628728
Dan Winship [Sun, 22 Nov 2009 21:10:48 +0000 (16:10 -0500)]
add a few missing introspection annotations
Dan Winship [Tue, 17 Aug 2010 12:28:33 +0000 (08:28 -0400)]
2.31.90
Dan Winship [Sun, 15 Nov 2009 17:05:16 +0000 (12:05 -0500)]
Try all IP addrs associated with a hostname if the first fails
Make SoupAddress keep track of multiple IP addresses and implement
GSocketConnectable, and make SoupSocket use GSocketClient (and thus
GSocketAddressEnumerator) to connect, so that it tries all addresses
associated with the SoupAddress.
In particular, this fixes the bug where if a host has both IPv4 and
IPv6 addresses, and glibc thinks you have IPv6 connectivity, but you
don't really, that libsoup can't connect to that host.
(Using GSocketClient rather than using GSocketAddressEnumerator
directly is a bit heavyweight for SoupSocket, since we don't actually
want to have a GSocket or GSocketConnection. But eventually, we will
be porting SoupSocket to use GSocket, and this code is a stepping
stone to that. Also, letting GSocketClient do the looping over the
addresses simplifies SoupSocket.)
(This change also silently breaks connection timeouts unless you have
the very latest git glib.)
https://bugzilla.gnome.org/show_bug.cgi?id=526321
Dan Winship [Wed, 11 Aug 2010 15:54:05 +0000 (11:54 -0400)]
Fix auto-Accept-Language in locales that use "," for decimals
noticed in related epiphany code in bug 602547
Dan Winship [Mon, 2 Aug 2010 21:38:15 +0000 (17:38 -0400)]
2.31.6
Dan Winship [Sat, 31 Jul 2010 08:32:51 +0000 (10:32 +0200)]
soup-message-io: fix wrong-Content-Length logic
Previously we just ignored Content-Length if the server specified
"Connection: close", which does the right thing if the specified
Content-Length is too large, but fails if the Content-Length is
correct but the server "forgets" to close the connection. Fix this so
that we always stop reading once we get to the expected
Content-Length, but we also cope with the connection closing before
that point.
https://bugzilla.gnome.org/show_bug.cgi?id=611481
Yaakov Selkowitz [Wed, 16 Jun 2010 04:10:13 +0000 (23:10 -0500)]
Do not dist gir_DATA
GIR files contain a shared-library attribute which varies per platform,
and therefore must not be disted; see bug 621611 for rationale.
https://bugzilla.gnome.org/show_bug.cgi?id=621727
Dan Winship [Fri, 30 Jul 2010 12:50:36 +0000 (14:50 +0200)]
Put the hostname into the reason_phrase with network errors
eg, "Cannot resolve hostname (sadfasdfasdfasdf.com)"
This can be useful for debugging.
https://bugzilla.gnome.org/show_bug.cgi?id=623274
Dan Winship [Fri, 30 Jul 2010 14:28:30 +0000 (16:28 +0200)]
SoupProxyResolverStatic: port to newer SoupProxyURIResolver API
Dan Winship [Fri, 30 Jul 2010 12:56:00 +0000 (14:56 +0200)]
SoupSessionAsync: don't stall on proxy resolution error
Dan Winship [Wed, 28 Jul 2010 12:51:06 +0000 (14:51 +0200)]
SoupSessionAsync: make soup_session_cancel_message behave stupidly again
The session rewrites changed cancellation in async sessions to happen
asynchronously. This makes more sense than the old behavior, but can
cause problems with code that previously assumed it could
cancel-and-forget (and it wasn't actually an intentional change
anyway). So revert that.
https://bugzilla.gnome.org/show_bug.cgi?id=625406
Dan Winship [Tue, 29 Jun 2010 13:43:20 +0000 (09:43 -0400)]
Disable TLS 1.2 in addition to 1.0 and 1.1
Due to bug 581342 we want to only negotiate SSL 3.0. Previously we
were telling gnutls to not do TLS1.0 or TLS1.1, but that means with
newer versions of gnutls that support TLS1.2 it would try to negotiate
that instead and generally fail. Fix that by disabling TLS1.2 too
(which works fine even with gnutls versions that don't support TLS1.2
yet).
https://bugzilla.gnome.org/show_bug.cgi?id=622857
Dan Winship [Sat, 12 Jun 2010 00:26:23 +0000 (20:26 -0400)]
Fix redirects-to-different-hosts to not reuse the same connection
We were mistakenly reusing the same SoupConnection for a queue item
even if the message got redirected to a different host. Fix that. The
particular fix here (always force it to re-run
soup_session_get_connection() after a redirect) is a bit
over-aggressive, but this will get rewritten when we eventually fix
connection pooling in the proxy case.
Also, add a test to tests/redirect-test for this.
Dan Winship [Fri, 11 Jun 2010 19:26:53 +0000 (15:26 -0400)]
Allow calling soup_test_server_new() multiple times in test programs
Dan Winship [Thu, 10 Jun 2010 00:06:51 +0000 (20:06 -0400)]
soup_date_new_from_string: fix an out-of-bounds memory access
soup_date_new_from_string() would sometimes read off the end of the
string, but tests/date didn't catch this even under valgrind, because
all of the strings were compile-time constants and they got laid out
in a way that never triggered a read of invalid memory. So tweak the
test to g_strdup each string before parsing it, which will let
valgrind notice if we read outside its bounds in the future.
https://bugzilla.gnome.org/show_bug.cgi?id=620288
Dan Winship [Wed, 9 Jun 2010 18:34:50 +0000 (14:34 -0400)]
Fix connection states so in-use connections don't get pruned
Since connections are assigned to messages before they're connected
now, they should go to IN_USE after connecting, not IDLE. Otherwise
the idle-connection-closing code might close connections that existing
queue items were planning to use.
Dan Winship [Wed, 9 Jun 2010 17:33:10 +0000 (13:33 -0400)]
SoupSession: tweak cancellation again
After the too-many-redirects merge, redirect-test was leaking a queue
item, because cancelling a message at the point it does resulted in
the completion_cb never running. Fix soup_session_cancel_message()
(again) to not have this problem.
Dan Winship [Wed, 9 Jun 2010 17:27:03 +0000 (13:27 -0400)]
Fix a signal-handler leak in the too-many-redirects patch
Dan Winship [Thu, 3 Jun 2010 19:23:22 +0000 (21:23 +0200)]
SoupSession: continue reorg, remove remaining signal-based flow control
Make the sessions more state-machiney and fix up
soup_session_cancel_message() by having it set the message to the
(new) FINISHING state and letting the session implementation pick it
up from there. Remove the remaining "finished" and "restarted" signal
handlers; move the base session's "finished" handler to
"soup_session_unqueue_item" and call that from the right places in the
subclasses.
Dan Winship [Wed, 2 Jun 2010 10:05:46 +0000 (12:05 +0200)]
SoupSession: bind connections to messages sooner, improve conn failure case
Rather than having connections be independent until a message is
actually sent on them, assign them to queue items when they're created
for them. This means that, eg, if another connection frees up while
we're waiting for the first connection to connect, that we can't take
advantage of that. But it simplifies a ton of other things.
Also, if a new connection fails, don't fail every other message on
that host; let them succeed or fail independently. When doing a
CONNECT though, keep track of the queue item that necessitated it so
we can error that out if the CONNECT fails, just like we would if the
actual soup_connection_connect() had failed.
https://bugzilla.gnome.org/show_bug.cgi?id=619633
Dan Winship [Sat, 29 May 2010 14:37:34 +0000 (16:37 +0200)]
Add SoupMessageQueueItemState, remove SoupMessageIOStatus
SoupMessageIOStatus was always really more about the session than the
message. (SoupServer I/O didn't use it at all.) Replace it with a new
SoupMessageQueueItemState, on the queue item rather than the message.
Dan Winship [Sat, 29 May 2010 13:24:51 +0000 (15:24 +0200)]
soup-message-io: use SoupMessageQueueItems and add SoupMessageCompletionFn
push SoupMessageQueueItem down into soup-message-io, and end the I/O
process by calling a completion callback rather than emitting the
SoupMessage signals directly, which gives the session greater control.
Dan Winship [Sat, 29 May 2010 13:05:55 +0000 (15:05 +0200)]
Misc SoupConnection API improvements
Use GCancellables in connect_async/connect_sync (plus updates needed
for behavior changes caused by that). Make
soup_connection_disconnect() always emit the signal if the connection
state wasn't already DISCONNECTED, even if it wasn't yet connected
before. Remove some dead code in clear_current_request. Remove some
more-or-less redundant arguments in soup_message_send_request().
Dan Winship [Sat, 29 May 2010 11:43:34 +0000 (13:43 +0200)]
Make SoupConnection warn if disposed "incorrectly"
Connections should always be disconnected and message-less by the time
they are disposed. Likewise, their SoupSockets should be disconnected
and not in the middle of an async connect when they're freed (though
we can't make the sockets warn unconditionally, since other SoupSocket
users external to libsoup may still be doing it "wrong", since that has
historically been a supported part of the API)
Fix some problems revealed by these warnings. Also, remove some weak
refs and replace others by hard refs as needed. Remove unnecessary
SoupSessionAsyncTunnelData type.
Dan Winship [Sat, 29 May 2010 14:14:33 +0000 (16:14 +0200)]
tests/auth-test: clean up a bit and fix async auth test logic
Ensure that we're not aborting the session until after all the
messages finish. (Previously it worked by coincidence, but upcoming
changes tweak things enough that it would fail otherwise.)
Dan Winship [Thu, 3 Jun 2010 08:37:39 +0000 (10:37 +0200)]
soup-status.c: #include config.h to get HAVE_SSL define
Now SOUP_STATUS_SSL will have the the reason phrase "SSL handshake
failed" instead of "SSL support not available" when SSL was available.
Dan Winship [Wed, 9 Jun 2010 13:35:25 +0000 (09:35 -0400)]
Fix (transfer) annotations on some non-transfer-full getters
https://bugzilla.gnome.org/show_bug.cgi?id=621021
Dan Winship [Mon, 7 Jun 2010 18:17:36 +0000 (20:17 +0200)]
sniffing-test: be robust against slightly-broken jhbuild configs
previously if you jhbuilt glib and libsoup, but not shared-mime-info,
then this test would fail. Fix that.
https://bugzilla.gnome.org/show_bug.cgi?id=612706
Dan Winship [Mon, 7 Jun 2010 12:57:52 +0000 (14:57 +0200)]
SoupProxyResolverGNOME: improve behavior with more-recent libproxy
SoupProxyResolverGNOME had a nasty workaround for the fact that
libproxy 0.2 would crash inside GConf if called from a thread other
than the main one. Unfortunately, the workaround involves leaking
proxy settings into the environment, which is particularly bad since
gnome-panel uses libsoup (via libgweather via intlclock) and so the
environment can become polluted for everything started from the panel.
In libproxy 0.3 and later, the workaround is no longer necessary, so
remove it if libproxy is new enough.
https://bugzilla.gnome.org/show_bug.cgi?id=603285
José Millán Soto [Fri, 4 Jun 2010 11:46:31 +0000 (13:46 +0200)]
SoupSession: count redirections and cancel the message after too many
and add a test to redirect-test
https://bugzilla.gnome.org/show_bug.cgi?id=604383
Dan Winship [Tue, 25 May 2010 00:00:02 +0000 (20:00 -0400)]
2.31.2