Sometimes the poll duration in the /socket/timed_wait test is slightly lower
than the requested 100000, causing failures like
ERROR:/build/buildd/glib2.0-2.33.2/./gio/tests/socket.c:619:test_timed_wait:
assertion failed (poll_duration > = 100000): (99240 >= 100000)
FAIL
Adjust the test to also allow some jitter in the "too small" direction, similar
to the already existing span for "slightly too large".
https://bugzilla.gnome.org/show_bug.cgi?id=678881
In general, code using g_slist_delete_link() is broken, because it
potentially requires an O(n) traversal. Just switch to GList in this
case.
The performance hit here was exacerbated by the fact that we were
holding a mutex that needed to be accessed by all threads.
https://bugzilla.gnome.org/show_bug.cgi?id=678576
make sure the proxy threads are in the "waiting for a connection"
state when we do the final cleanup, or else there are race conditions
involving which thread processes the GCancellable cancellation first.
The bash-completion code nowadays expects completion files to
be installed in /usr/share/bash-completion/completions, and
expects them to be named like the command they are completing
for.
https://bugzilla.gnome.org/show_bug.cgi?id=677782
-gconverterinputstream.c: Avoid GCCism by not using non-standard pointer
arithmetic on void*, but do a cast to char * as that seems to be what the
variable was used for.
-gtestdbus.c: Don't include unistd.h unconditionally, and use g_usleep()
instead of usleep(), as usleep() is not universally available.
The "-framework" linker flag takes a second word as a parameter. If
they are passed separated with whitespace, some flag-handling routines
may not know to keep the two words together as a single unit. Use
-Wl,, to pass multiple words without embedded whitespace.
https://bugzilla.gnome.org/show_bug.cgi?id=566994
Reading from a GConverterInputStream with both input_buffer and
converted_buffer non-empty would return bogus data (the data from
converted_buffer would essentially get skipped over, though the
returned nread reflected what the count would be if it hadn't been).
This was never noticed before because (a) it can't happen if all of
your reads are at least as large as either the internal buffer size or
the remaining length of the stream (which covers most real-world use),
and (b) it can't happen if all of your reads are 1 byte (which covers
most of tests/converter-test). (And (c) it only happens for some
converters/input streams.) But this was happening occasionally in
libsoup when content-sniffing a gzipped response, because the
SoupContentSnifferStream would first read 512 bytes (to sniff), and
then pass through larger reads after that.
Fixed and added a test to converter-test.
https://bugzilla.gnome.org/show_bug.cgi?id=676478
When the application is using its D-Bus backend, it is useful to be able
to export extra D-Bus objects at the right time, i.e. *before* the application
tries to own the bus name. This is accomplished here by adding a hook
in GApplicationClass for this; and a corresponding hook that will be called
on unregistration to undo whatever the register hook did.
Bug #675509.
I didn't do this comprehensively, since there's a lot of it, mainly
due to the GDBus object manager stuff, but anyone trying to use
that would fail fast due to lack of the gdbus code generator.
My main goal was to get API additions to existing classes like
g_data_input_stream_read_line_utf8(), as well as the lower level new
API like glib-unix.h.
https://bugzilla.gnome.org/show_bug.cgi?id=676816
Using a caller-supplied buffer for g_input_stream_read() doesn't
translate well to the semantics of many other languages, and using a
non-refcounted buffer for read_async() and write_async() makes it
impossible to manage the memory correctly currently in
garbage-collected languages.
Fix both of these issues by adding a new set of methods that work with
GBytes objects rather than plain buffers.
https://bugzilla.gnome.org/show_bug.cgi?id=671139
Rather than depending on the host's DNS configuration to properly
return an error for a non-existent hostname, just substitute in
a dummy GResolver implementation that does it for us.
GFile doesn't handle some "real" URIs, so check if there's a default
handler for the URI scheme first, and only use g_file_new_for_uri()
and g_file_query_default_handler() if not. Eg, this fixes the case of
opening http URIs with "%2F" in the path.
https://bugzilla.gnome.org/show_bug.cgi?id=666386
This essentially adds an accessor for the MimeType field in desktop files,
to retrieve the list of all mime types supported by an application.
The interface though is part of GAppInfo, so it could be implemented
in the future by other backends.
https://bugzilla.gnome.org/show_bug.cgi?id=674111
The logic here is pretty twisted, but basically we were leaking a ref
for each non-existent parent. The clearest way to fix this was to
move to more explicit refcounting logic; when a variable is pointing
to an object, it holds a ref.
https://bugzilla.gnome.org/show_bug.cgi?id=675446
Provide public access to the GDBusConnect and object path that
GApplication is using. Prevents others from having to guess these
things for themselves based on the application ID.
https://bugzilla.gnome.org/show_bug.cgi?id=671249
g_output_stream_write_async() was not initializing the newly-added
members of the WriteData structure, causing various problems.
Also, g_input_stream_read_async() was now leaking its cancellable. Fix
that as well.
https://bugzilla.gnome.org/show_bug.cgi?id=674612
If the launch context is a GAppLaunchContext, and not a
GdkAppLaunchContext, then g_app_launch_context_get_display will return
NULL because the get_display virtual method is undefined. The DISPLAY
might still be inherited from the parent process, in which case
overwriting it with NULL breaks the launch.
This is a regression introduced in:
de834bed30
Fixes: https://bugzilla.gnome.org/672786
Commit f084b60377 incorrectly set
DIST_SUBDIRS for the toplevel Makefile.am. In general actually we
don't need to set it, because modern automake automatically sets
it by looking at conditionals for SUBDIRS.
Tested-by: Rico Tzschichholz <ricotz@t-online.de>
https://bugzilla.gnome.org/show_bug.cgi?id=667806
If all members of GSocketFamily are supported on the platform, then
all of its values will be positive, and so the enum might become
unsigned, in which case testing for "family < 0" might cause warnings.
But we want to return an error if family == 0 (aka
G_SOCKET_FAMILY_INVALID) anyway, so just tweak the test accordingly.
https://bugzilla.gnome.org/show_bug.cgi?id=674592
gdbus-daemon-generated.[ch] failed to build because it depended
on gdbus-2.0/codegen/gdbus-codegen which was build during the SUBDIRS part
of the build, however SUBDIRS are done *after* processing BUILT_SOURCES,
and these files are in BUILT_SOURCES.
The fix is simple, instead of running the gdbus-codegen code we
run the gdbus-codegen.in code, which works fine for uninstalled execution.
I also removed Makefile from the dependencies to avoid rebuilding the file
in tarballs, as Makefiles are written at configure time. We should be able to
ship the prebuilt files in the tarballs.
When running uninstalled
Add two new methods to GProxyAddress for recovering information about
the destination URI that the proxy was created for (and modify
GProxyAddressEnumerator to set that information when creating the
GProxyAddress).
In the async case, a failed DNS lookup was causing the proxy
resolution to bail out immediately, rather than just moving on to the
next potential proxy (which might not need us to do the DNS lookup
beforehand). Fix that.
This is mostly complete, sans support for activation. However, its
not as picky as the libdbus implementation in terms like validation
and limits checking, nor is it as tested.
Its can be useful to test gdbus if dbus-daemon is not availible, but
its main reason for existance is to implement a default session bus
on win32 so that e.g. GApplication is guaranteed to work.
If a GInputStream does not provide a read_async() implementation, but
does implement GPollableInputStream, then instead of doing
read-synchronously-in-a-thread, just use
g_pollable_input_stream_read_nonblocking() and
g_pollable_input_stream_create_source() to implement an async read in
the same thread. Similarly for GOutputStream.
Remove a bunch of existing read_async()/write_async() implementations
that are basically equivalent to the new fallback method.
https://bugzilla.gnome.org/show_bug.cgi?id=673997
Implement GPollableInputStream in GMemoryInputStream and
GConverterInputStream, and likewise implement GPollableOutputStream in
the corresponding output streams.
https://bugzilla.gnome.org/show_bug.cgi?id=673997
Move g_pollable_source_new() here from gpollableinputstream.c, add
g_pollable_source_new_full(), and add some new methods to do either
blocking or nonblocking reads depending on a boolean argument.
https://bugzilla.gnome.org/show_bug.cgi?id=673997
Make g_pollable_input_stream_read() and
g_pollable_output_stream_write() look a little bit more like the
non-pollable versions in terms of error handling, etc. Also, use the
read_fn and write_fn virtual methods directly rather than calling
g_input_stream_read()/g_output_stream_write(), to avoid problems with
re-entrancy involving the "pending" flag.
Also belatedly add single-include guards to the header files.
https://bugzilla.gnome.org/show_bug.cgi?id=673997
The loop was using a GConverterResult variable where it meant to use a
gssize, and since GConverterResult was ending up as an unsigned type,
this meant the (res < 0) check always failed.
Resources are always little endian, so the gvdb is byteswapped. When looking
up the value, it would return a new byteswapped variant, making the data
returned from do_lookup() invalid once that variant is unref'd. Since
byteswapping doesn't matter for the "ay" data anyway, just use
gvdb_table_get_raw_value() instead and only byteswap the length and flag
values.
https://bugzilla.gnome.org/show_bug.cgi?id=673409
* Add resolver functions for looking up DNS records of
various types. Currently implemented: MX, TXT, SOA, SRV, NS
* Return records as GVariant tuples.
* Make the GSrvTarget lookups a wrapper over this new
functionality.
* Rework the resolver test so that it has support for
looking up MX, NS, SOA, TXT records, and uses GOptionContext
https://bugzilla.gnome.org/show_bug.cgi?id=672944
This patch solves two problems:
First, it allows builders to optionally cut the circular dependency
between dbus and glib by disabling the modular tests (just like how
the tests can be disabled in dbus).
Second, the tests are entirely pointless to build if cross-compiling.
It also moves us slightly closer to the long term future we want where
the tests are a separate ./configure invocation and run against the
INSTALLED glib, not the one in the source tree. This would allow us to
run the tests constantly, not just when glib is built.
https://bugzilla.gnome.org/show_bug.cgi?id=667806
For quite some time the recommended usage of GSettings and dconf has
been to use paths like /org/gnome/example/. Use of /apps/ has spilled
over from GConf and is continuing to make its way into a number of
applications as they port.
glib-compile-schemas will now warn about these types of paths being
used. This generates a lot of noise, but hopefully it will reduce the
number of ported applications making this mistake.
Turns out libdbus doesn't send struct ucred credentials on linux, but
just relies on the SO_PEERCRED support. However, gdbus does send, and
expect to recieve a ucred credential. So, when libdbus talks to a
gdbus server the authentication fails to send the credentials.
We fix this by falling back to g_socket_get_credentials() if we don't
get any credential messages.
When presented with an array of empty arrays of 8-byte-aligned types,
GDBus would incorrectly apply the 8-byte alignment when reading back.
https://bugzilla.gnome.org/show_bug.cgi?id=673612
Signed-off-by: David Zeuthen <davidz@redhat.com>
D-Bus arrays are serialized as follows:
1. align to a 4-byte boundary (for the length)
2. uint32: the length of the serialized body in bytes
3. padding for the alignment of the body type (not included in the length)
4. the body.
Note that 3. is a no-op unless the body type is an 8-byte aligned type
(uint64, int64, double, struct, dict_entry), since you are always on a
4-byte boundary from aligning and writing the length.
So, an empty aax (that is, an array containing zero arrays of int64)
is serialized as follows:
1. align to a 4-byte boundary
2. length of the contents of this (empty) array, in bytes (0)
3. align to a 4-byte boundary (the child array's alignment requirement)
4. there is no body.
But previously, GDBus would recurse in step three to align not just for
the type of the child array, but for the nonexistent child array's
contents. This only affects the algorithm when the grandchild type has
8-byte alignment and the reader happened to not already be on an 8-byte
boundary, in which case 4 bytes were spuriously skipped.
https://bugzilla.gnome.org/show_bug.cgi?id=673612
Signed-off-by: David Zeuthen <davidz@redhat.com>
