| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
| |
| |
| |
| |
| |
| |
| | |
g_warn_if_fail() if the Content-Type is invalid, and impove the
annotations.
https://bugzilla.gnome.org/show_bug.cgi?id=686766
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
If a SoupMessage which hasn’t yet started to be received is passed
into soup_server_[un]pause_message(), an assertion will fail in
soup-message-io.c because the message doesn’t yet have any I/O data.
Document that passing such SoupMessages into those methods is naughty
and shouldn’t be done.
https://bugzilla.gnome.org/show_bug.cgi?id=710602
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Since automake's "make check" rule now logs the output of tests to a
log file, we might as well use maximum debugging, so that if something
fails, we have plenty of explanation why.
(We still parallelize parallelizable tests by default though, meaning
the output is not especially useful for those ones.)
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
g_input/output_stream_close() always mark their stream closed, even on
error, even if their cancellable is pre-cancelled. That means we have
to guarantee that soup_message_io_finished() gets called when a
SoupClientInputStream is closed, even if an error occurs while
closing.
https://bugzilla.gnome.org/show_bug.cgi?id=711260
|
| | |
|
| | |
|
| |
| |
| |
| |
| |
| |
| |
| | |
GTask needs to be unreffed after calling g_task_return_* when we
don't use async methods which take ownership of it (as
g_task_run_in_thread() does).
https://bugzilla.gnome.org/show_bug.cgi?id=711085
|
| |
| |
| |
| |
| |
| |
| | |
glib API documentation indicates:
« The returned string should be freed when no longer needed. »
https://bugzilla.gnome.org/show_bug.cgi?id=711085
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Otherwise we may attempt to remove a nonexistent quit loop from the
previous one. The second hunk here is the real fix, the first is just
being more defensive/consistent.
Reported-By: Kjartan Maraas <kmaraas@gnome.org>
Tested-By: Kjartan Maraas <kmaraas@gnome.org>
https://bugzilla.gnome.org/show_bug.cgi?id=711361
|
| | |
|
| |
| |
| |
| |
| |
| | |
When using soup_session_send() or soup_session_send_async(), propagate
underlying GErrors to the caller, rather than constructing fake
SOUP_HTTP_ERROR errors.
|
| |
| |
| |
| |
| |
| |
| | |
Port the SoupConnection methods to be gio-like, using
GAsyncReadyCallback for the async ones, and returning GErrors rather
than libsoup status codes. Add internal-gio-like SoupSocket connect
methods for the SoupConnection methods to use.
|
| |
| |
| |
| | |
and update SoupConnection to use them
|
| |
| |
| |
| |
| |
| |
| |
| | |
When closing the stream early, we should be skipping the read state
ahead to FINISHING, rather than BODY_DONE, so that got_body doesn't
get emitted (thus causing the connection to mark itself reusable).
https://bugzilla.gnome.org/show_bug.cgi?id=710494
|
| |
| |
| |
| | |
https://bugzilla.gnome.org/show_bug.cgi?id=710267
|
| | |
|
| | |
|
| | |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Change uri_decoded_copy() to strdup the input string *after* doing its
g_return_val_if_fail(), to avoid pointlessly confusing code analyzers.
Also, rename the function to soup_uri_decoded_copy(), since it's used
from soup-request-data.c too and it's been annoying me that it was
non-namespaced.
https://bugzilla.gnome.org/show_bug.cgi?id=709793
|
| | |
|
| |
| |
| |
| |
| |
| | |
One of connection-test's tests was failing (but only sporadically)
because it closed the stream without reading it, causing the stream to
be disconnected, while it was expecting the connection to stay open.
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
plain http or https
We can check only for "tls-certificate", once it's set
whether "ssl-cert-file" and "ssl-key-file" are set.
https://bugzilla.gnome.org/show_bug.cgi?id=709647
|
| |
| |
| |
| |
| |
| |
| |
| | |
If the user overrides the proxy or TLS configuration of a session at
construct time, we should avoid ever trying to resolve the default
extensions.
https://bugzilla.gnome.org/show_bug.cgi?id=708696
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Closing a SoupClientInputStream for a message that hadn't been
completely read was trying to read to the end of the message first.
Fix it to just cancel the read instead.
Also fix a few tests that were implicitly assuming the old behavior.
https://bugzilla.gnome.org/show_bug.cgi?id=695652
|
| |
| |
| |
| |
| | |
To ensure we don't get unwanted proxy settings, and so we don't get
warnings if the memory backend is the only one available.
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Messages sent via soup_session_queue_message() on a plain SoupSession
accidentally ended up using blocking I/O. Fix this (and also make
switching between sync and async ops during a streaming operation work
better).
https://bugzilla.gnome.org/show_bug.cgi?id=707711
|
| |
| |
| |
| |
| |
| |
| | |
From php_url_encode:
/* Allow only alphanumeric chars and '_', '-', '.'; escape the rest */
https://bugzilla.gnome.org/show_bug.cgi?id=708621
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
There were two issues. Firstly, when the automatic NTLM authentication
failed we were setting conn->state to SOUP_NTLM_FAILED. So although we
would emit the 'authenticate' signal and obtain a password, we'd never
actually *try* it. We have the SOUP_NTLM_SSO_FAILED state for this, which
makes it work.
Secondly, the resulting 'authenticate' signal we being emitted with the
'retrying' property set, even when it was the first time we actually asked
for a password. This was confusing Evolution (bug 703181 comments 1-3).
Fix the latter problem by making soup_auth_ntlm_is_connection_ready()
return TRUE when the state is SOUP_NTLM_SSO_FAILED. That makes
auth_got_headers() do the right thing, and Evolution is happy again.
With this fixed, Evolution will now automatically try to connect to an
Exchange server using automatic NTLM authentication, and will fall back
to asking for a password only if it actually needs to.
|
| |
| |
| |
| |
| |
| |
| | |
We need to clear priv->reusable when we send a request, *even* if we're
only resending a request that was already tried once (as with NTLM
authentication). Otherwise, cancellation of the pending, restarted,
request will not properly close the connection as it should.
|
| | |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
On slow old machines, the TLS handshake when making an https request
can take longer than 1 second, which meant that even the "fast" https
requests were hitting the 1-second timeout and failing.
Fix this by making the timeouts longer. However, we don't want to do
that unless absolutely necessary, since the test has to hit this
timeout multiple times, and the test already takes 10 seconds to run
with the 1-second timeout. So check how long it takes to make a single
https request, and only slow down the rest of the tests if it looks
like it will be needed.
Also fix an unrelated bug, where the "session did/didn't close
connection" check could fail if the same memory location got used for
the first and second sockets.
https://bugzilla.gnome.org/show_bug.cgi?id=660581
|
| | |
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Add :http-aliases and :https-aliases properties to SoupServer, to
allow it to handle requests like:
GET daap://host:port/path HTTP/1.1
https://bugzilla.gnome.org/show_bug.cgi?id=703694
|
| | |
|
| | |
|
| |
| |
| |
| |
| |
| | |
If doing a read_line()/read_until() in a subclass, we need to use the
subclass's read implementation rather than just going directly to the
base stream.
|
| |
| |
| |
| |
| |
| |
| |
| | |
When closing a chunked stream, we need to write the final 0-length
chunk. However, we can only do that if we successfully wrote the
previous chunk.
https://bugzilla.gnome.org/show_bug.cgi?id=703297
|
| |
| |
| |
| | |
https://bugzilla.gnome.org/show_bug.cgi?id=691840
|
| |
| |
| |
| | |
https://bugzilla.gnome.org/show_bug.cgi?id=706338
|
| |
| |
| |
| |
| |
| |
| | |
The refcounting test had a race condition. Fix it by just waiting for
the stream to be destroyed, rather than asserting it already has been.
https://bugzilla.gnome.org/show_bug.cgi?id=698305
|
| | |
|
| |
| |
| |
| |
| |
| |
| | |
Don't include the scope ID in the "Host" header.
Make SoupURI correctly parse both correct and incorrect scope IDs, and
add checks for that in tests/uri-parsing.
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
SoupAddress's GSocketAddressEnumerator implementation was passing the
wrong GMainContext to soup_address_resolve_async(), although it only
ended up affecting the address-is-already-resolved case. Fix this.
Also, remove the async_lookups list from SoupAddressPrivate, and just
make multiple calls to the resolver if there are multiple calls to
soup_address_resolve_async() at once, since that code didn't handle
different GMainContexts correctly.
|
| |
| |
| |
| |
| |
| |
| | |
Use GNetworkAddress to try to parse IPv6 addresses with scope IDs,
rather than just letting the scope ID get dropped.
Part of https://bugzilla.gnome.org/show_bug.cgi?id=669724
|
| | |
|
| | |
|