Here is a WIP of the new GSettings compiler based on GMarkupReader.

This is actually a utility program that dumps out a document by way of
GMarkupReader.

Add GMarkupReader which is a new GMarkup-style parser that reads from a
GInputStream and operates as a sort of iterator that gets advanced to
each token in sequence and then inspected.

The API makes it quite nice to write parsers where the function nesting
(in C) follows the nesting of the elements in the file being parsed.

It's also possible to use this new class to write async or
partially-async parsers in which the stream is advanced until reading
would block (via GPollableInputStream) and only then is an async call
made.

Expose some of the guts of GMarkupParseContext for use via the
glib-private interface, from GIO.

There are some functionality changes in this commit that should probably
also be split out.....

20e709d8 fixed some issues in
configure.ac but also introduced a typo.  Fix that.

Alessandro Di Federico authored Jul 25, 2014 and Allison Karlitskaya committed Jul 26, 2014

More of the same as 9362a1b2.

https://bugzilla.gnome.org/show_bug.cgi?id=733715

It turns out that this bug actually would (sometimes) impact any sort of
fixed-sized array with an alignment requirement of 8 due to incorrectly
counting the alignment inserted between the (aligned 4) array length and
the actual data.

Fix this properly and remove the exception for doubles.

https://bugzilla.gnome.org/show_bug.cgi?id=732754

We're using a precondition in the middle of the function, and if we
hit it, we leak the closure.

Let's allocate the closure per path; this allows us to allocate it
before path-specific preconditions, and better avoids a pointless
malloc/free pair in the unhandled case.

https://bugzilla.gnome.org/show_bug.cgi?id=733576

Take the simple slow path in this case. Encoding a double
takes special precautions as you can see in append_value_to_blob()
and friends.

https://bugzilla.gnome.org/show_bug.cgi?id=732754

https://bugzilla.gnome.org/show_bug.cgi?id=732754

This reverts commit 7269d753.

Adding G_STATIC_ASSERT() into a header file caused compilation
problems with at least one app (Anjuta). Reverting to keep
GNOME continuous testing running.

https://bugzilla.gnome.org/show_bug.cgi?id=730932

This should always be true.

https://bugzilla.gnome.org/show_bug.cgi?id=733576

We know the group exists since we enumerated it.  Should silence a
static analysis warning.

https://bugzilla.gnome.org/show_bug.cgi?id=733576

Just ignore the return value, since we're checking contents != NULL.

https://bugzilla.gnome.org/show_bug.cgi?id=733576

We're intentionally ignoring the value here.  Pacifies static
analysis.

https://bugzilla.gnome.org/show_bug.cgi?id=733576

Spotted by static analysis.

https://bugzilla.gnome.org/show_bug.cgi?id=733576

Spotted by static analysis.

https://bugzilla.gnome.org/show_bug.cgi?id=733576

I don't believe any real app would hit this, but we do leak
if it occurs.

Spotted by static analysis.

https://bugzilla.gnome.org/show_bug.cgi?id=733576

Discovered by static analysis.

https://bugzilla.gnome.org/show_bug.cgi?id=733576

Not a real bug, but will quiet the analysis.

https://bugzilla.gnome.org/show_bug.cgi?id=733576

configure.ac assumes this.

Bug: https://bugzilla.gnome.org/show_bug.cgi?id=730932

This code assumes that int is exactly 4 bytes, and that pointers
are either 4 or 8 bytes, on platforms with __ATOMIC_SEQ_CST.
In practice this is going to be true.

Bug: https://bugzilla.gnome.org/show_bug.cgi?id=730932

This is a best-effort approach to preventing SIGPIPE emissions on Darwin
and iOS, where they continue to be intercepted by the Xcode debugger
even if SIG_IGN prevents them crashing the program.

This is similar to the existing code which sets MSG_NOSIGNAL on all
send() calls. MSG_NOSIGNAL doesn't exist on Darwin though.

Based on a patch from Philip Withnall.

https://bugzilla.gnome.org/show_bug.cgi?id=728730

Ensure that the input string is non-NULL, as inet_pton() does not like
NULL strings.

https://bugzilla.gnome.org/show_bug.cgi?id=733345

Remove checks for NULL before g_free().

https://bugzilla.gnome.org/show_bug.cgi?id=733156

Ignacio Casal Quinteiro authored Jul 14, 2014 and Matthias Clasen committed Jul 14, 2014

Since we are getting passed Unicode values these global vars
might not have the correct value. Instead always get the wide arguments
and convert them to utf8 to use them.

https://bugzilla.gnome.org/show_bug.cgi?id=733146

Gergely Polonkai (naurYl) authored Jul 12, 2014 and Matthias Clasen committed Jul 14, 2014

https://bugzilla.gnome.org/show_bug.cgi?id=733084

SyncEvolution Nightly Testing authored Jul 10, 2014 and Matthias Clasen committed Jul 14, 2014

When establishing a D-Bus connection failed, g_dbus_object_manager_client_finalize()
calls g_object_ref(manager->priv->connection) when that pointer is NULL,
which is considered and logged as error by glib.

https://bugzilla.gnome.org/show_bug.cgi?id=732984

This fix is a bit of a blind guess, I couldn't really discern
from recent commits whether this is an expected behavior change
or not.