We need to do this because constructors run before main() and
thus before any call to g_mem_set_vtable, making it impossible to
use that function if constructors call g_malloc.
We do this by making the constructors just register the static data
for lazy registration, doing the lazy registration when using
the global resource set.
With this we're not longer exporting the constructor headers, which means
we're not tying ourselves to a macro that might need special tweaking on
a compiler-by-compiler basis.
It's important to have strict rules for handling of whitespace in
translated strings in GSettings schema files so that the tools
extracting the messages will end up with the same messages as the
runtime calling gettext().
The rules are designed to be simple and unambiguous yet cover most
normal uses in a convenient way.
Those rules are as follows (with rationale):
- for <default> tags, the text content has its leading and trailing
whitespace stripped off, but internal whitespace is not modified in
any way.
This allows for slightly more flexible use of whitespace without
causing that whitespace to appear in the strings for translation.
- for <summary> and <description> tags, the content is split into
paragraphs. Paragraphs are separated by two or more sequential
newline characters. Each paragraph has its leading and trailing
whitespace removed and all other whitespace is normalised to a
single ascii space character. Finally, the paragraphs are rejoined,
inserting exactly two newlines between them.
This allows for longer explanations (particularly in the description
tag) using a natural format that, when normalised, will display
nicely in toolkits.
This patch implements the rules for <default> tags. The schema compiler
currently ignores <summary> and <description> tags.
First, correct a rather dubious case of accessing a GSettingsSchemaKey
after clearing it. This was technically okay because only the key name
was accessed (and it is not owned by the struct) but it looks very
wrong.
Second, have g_settings_backend_write() sink the passed in GVariant*.
Not all backends get this right, and I'm starting to like the pattern of
virtual function wrappers being responsible for sinking the parameters
that they are documented as consuming.
GValueArray was deprecated in bug 667228 and since we never change the
size of the array, it was kinda dumb to just GValueArray in the first
place.
https://bugzilla.gnome.org/show_bug.cgi?id=667228
Signed-off-by: David Zeuthen <davidz@redhat.com>
GKeyFile supports empty files and also supports loading from the string
"", but will fail with a critical if you try:
- explicit length == 0
- data == NULL
length == 0 should always be valid, and data == NULL should be valid in
the case that length == 0, so add some testcases for those and fix the
code up to allow them.
https://bugzilla.gnome.org/show_bug.cgi?id=668756
This is needed for thread-safety ... yes, it would have been better to
make get_object() return a full reference and have something like a
peek_object() method return a borrowed reference for C convenience
(only a single vfunc would have been needed). But such an ABI break is
too late now...
Signed-off-by: David Zeuthen <davidz@redhat.com>
And remove a comment about Windows in the fallback implementation that
no longer applies, since there's now a separate Windows-specific
implementation.
It's hardly useful to bloat the resource data with blanks intended only
for human readability, so add a preprocessing option that uses xmllint --noblanks
to strip these.
Bug #667929.
