summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAgeFilesLines
* session: handle request cancellation earlierHEADmastercarlosgc/early-cancellationCarlos Garcia Campos2023-05-083-27/+139
| | | | Check if the message was cancelled before the request is sent.
* 3.4.23.4.2Patrick Griffis2023-04-292-1/+5
|
* Revert "session: handle request cancellation earlier"Patrick Griffis2023-04-292-29/+1
| | | | This reverts commit 517eb93f33e3d5dcf7ba059608114fc9e36b62af.
* Revert "session: do not finish an already finished item"Patrick Griffis2023-04-291-1/+1
| | | | This reverts commit b9f4f0367704e4d7f6004645c24887a979014f10.
* Revert "connection: a connection is disconnected state should not be reusable"Patrick Griffis2023-04-291-2/+1
| | | | This reverts commit a15107a70b4b2b0b5bda241f037d09af403db96e.
* connection: a connection is disconnected state should not be reusableCarlos Garcia Campos2023-04-211-1/+2
| | | | This might fix #342
* 3.4.13.4.1Patrick Griffis2023-04-202-1/+7
|
* lib: Add names to various GSourcesPhilip Withnall2023-04-137-1/+70
| | | | | | | | | | | | This helps in identifying what’s attached to a `GMainContext` when debugging runtime problems with libsoup. The names added in `soup-misc.c` are fairly rough; if they turn out to not be specific enough, those methods could be changed to accept a `const gchar *name` argument so the caller can provide a more specific name. Signed-off-by: Philip Withnall <pwithnall@endlessos.org>
* lib: Add g_task_set_source_tag() everywherePhilip Withnall2023-04-138-0/+16
| | | | | | | This makes it easier to identify what a `GTask` instance is when debugging a running process. Signed-off-by: Philip Withnall <pwithnall@endlessos.org>
* tests: Fix warnings about unused results from g_string_free()Philip Withnall2023-04-132-6/+15
| | | | | | | | | | | | | | | | GLib 2.76 emits a warning if `g_string_free (_, FALSE)` is used without using its return value, as it typically signifies a leak: ``` warning: ignoring return value of ‘g_string_free_and_steal’ declared with attribute ‘warn_unused_result’ [-Wunused-result] ``` In this case, the libsoup example and test code did *not* leak because it had already stolen the string data from `GString`. That’s a bit of a layering violation (the string data briefly has two owners), so rearrange the code to fix that and silence the warning. Signed-off-by: Philip Withnall <pwithnall@endlessos.org>
* session: do not finish an already finished itemCarlos Garcia Campos2023-04-111-1/+1
| | | | This can happen now that we are handling the cancellation earlier.
* session: handle request cancellation earlierCarlos Garcia Campos2023-04-082-1/+29
| | | | Check if the message was cancelled on every queue state loop iteration.
* Update Bulgarian translationAlexander Shopov2023-04-071-3/+3
|
* Update Bulgarian translationAlexander Shopov2023-03-311-2/+2
|
* Fix strict-prototype warningPatrick Griffis2023-03-171-1/+1
|
* Fix regression breaking HTTP/2 on platforms with unsigned charPatrick Griffis2023-03-171-1/+1
|
* 3.4.03.4.0Patrick Griffis2023-03-172-1/+7
|
* soup-listener: Plug a leakAlessandro Bono2023-02-241-0/+2
| | | | Fixes: 25e96cad913cde650ca56ff76c3924fd97eb7125
* Add a missing nullable annotationLukáš Tyrychtr2023-02-201-1/+1
| | | | Fixes #335
* Fix handling of sniffed dataPatrick Griffis2023-02-151-13/+22
| | | | Closes #334
* Post-release bumpPatrick Griffis2023-02-121-1/+1
|
* Prepare 3.3.13.3.1Patrick Griffis2023-02-121-0/+6
|
* http1: fix crash when skipping respone stream with chunked encondingCarlos Garcia Campos2023-02-082-2/+71
| | | | | | | When skipping the response body the given buffer is NULL, which is not handled for chunked enconding when reading the trailers. Fixes #323
* Do not wait for the next loop iteration to unqueue async itemsCarlos Garcia Campos2023-02-038-99/+14
| | | | | | This way we release the connection earlier, since other requests might be waiting for the connection. We can also remove manual loop iterations in the tests after a request is done.
* Update Belarusian translationVasil Pupkin2023-01-231-12/+11
|
* cookie-jar: Fix valid Secure cookies being rejectedPatrick Griffis2023-01-232-6/+13
| | | | | The documentation for soup_cookie_jar_add_cookie_full() states NULL uris are always treated as a secure origin.
* Post-release bumpPatrick Griffis2023-01-201-1/+1
|
* Release 3.3.03.3.0Patrick Griffis2023-01-201-0/+33
|
* websocket-test: Fix minor warningsPatrick Griffis2023-01-201-3/+1
|
* soup-message: Publicize API to force HTTP/1 version for sendingMilan Crha2023-01-202-0/+44
| | | | | | | | This way also the libsoup library users can force HTTP/1 version only for some messages, not for the whole process (through the environment variable). Closes https://gitlab.gnome.org/GNOME/libsoup/-/issues/261
* soup-connection-manager: Read 'SOUP_FORCE_HTTP1' environment variable only onceMilan Crha2023-01-201-1/+5
| | | | | | The environment variables usually do not change during runtime, it's unexpected even for this semi-debugging variable, thus let it be read only once and the result being used repeatedly.
* connection-manager: free unused hosts always in the session contextCarlos Garcia Campos2023-01-201-12/+35
| | | | And with the connection manager mutex locked.
* cookies: Only consider space and tab as whitespaceSiwei Li2023-01-191-9/+7
| | | | | | The cookies spec defines whitespace as that. Previously we handled everything `isspace()` considered whitespace such as newlines which is incorrect.
* cookie-jar: Reject cookies with control characters in name or valueSiwei Li2023-01-193-4/+92
|
* cookies: Add support for cookie prefixesSiwei Li2023-01-162-0/+80
| | | | | | | | | | This is a spec that Firefox and Chrome implement: https://datatracker.ietf.org/doc/html/draft-ietf-httpbis-cookie-prefixes-00 Closes #349 Closes #326 Co-authored-by: Patrick Griffis <pgriffis@igalia.com>
* CI: Generate coverage reportsPatrick Griffis2023-01-163-3/+7
| | | | Closes #290
* docs: Run gtkdoc-rebase to fix relative glib linksPatrick Griffis2023-01-171-0/+1
|
* ci: Add gtk-doc to imagePatrick Griffis2023-01-173-2/+3
|
* docker: Workaround wsaccel 0.6.4 not installingPatrick Griffis2023-01-171-1/+2
|
* docs: Add robots.txtPatrick Griffis2023-01-172-1/+6
|
* CI: Fix showing test results in gitlabPatrick Griffis2023-01-161-1/+5
|
* docs: Fix dependency namesCorey Berla2023-01-161-1/+1
| | | | | Dependency names need to match the namespace name exactly (including capitalization) otherwise they won't be shown as dependencies.
* SameSite=None cookies should be rejected unless the Secure attribute is set.Siwei Li2023-01-142-3/+13
|
* message: fix critical when trying to create a message from invalid URIGuillaume Desmottes2023-01-121-1/+1
| | | | | | | | | soup_message_new_from_uri() raises a critical if SOUP_URI_IS_VALID returns FALSE. This macro also checks if the URI host is not an empty string, which was not checked in the caller code. Fix critical when running such GStreamer pipeline: gst-launch-1.0 souphttpsrc location=http:///not-there.ogg ! fakesink
* message: add SoupMessage::got-body-data signalCarlos Garcia Campos2023-01-116-16/+90
| | | | | | It can be used to monitor the progress of the read operation. Fixes #319
* soup_uri_copy: Don't retain default ports when copyingPatrick Griffis2023-01-092-1/+36
| | | | Fixes #322
* Update Belarusian translationVasil Pupkin2023-01-091-91/+105
|
* soup-connection-auth: Add thread safety around 'conns' hash tableMilan Crha2023-01-021-3/+13
| | | | | | | The GHashTable is not thread safe, thus add a thread safety around it, to avoid issues when it's accessed from multiple threads at the same time. Closes https://gitlab.gnome.org/GNOME/libsoup/-/issues/317
* session: add soup_session_send_and_splice() and ↵Carlos Garcia Campos2022-12-143-95/+256
| | | | | | | soup_session_send_and_splice_async() This is convenient API for reading the request body stream and writing it to an output stream.
* server-connection: fix dep cycle and connection leakBiagioFesta2022-12-131-0/+1
| | | | | | | | | The initial message contains a cycle-reference to the connection. We need to decrement the reference in order to properly clean up the connection. Before this patch, it was possible SoupServerConnection-finalize was never called. This was causing a memory leak (and socket FD leak).