A few applications such as gnome-music load the GIRepository typelib
and use it to adjust their search paths.
GLib 2.79.x now provides libgirepository-2.0.so.0 (GIRepository-3.0),
but each OS distribution is likely to have a transitional period during
which GLib's libgirepository-2.0.so.0 has become available, but bindings
like PyGI and gjs are still linked to gobject-introspection's
libgirepository-1.0.so.1 (GIRepository-2.0).
During this transitional period, interpreted languages that load the
GIRepository namespace could get the "wrong" version, which will result
in adjusting a search path that will not actually affect the language
binding's typelib lookup, and could also lead to symbol and type-system
conflicts.
We can avoid this collision by making GLib's GIRepository library refuse
to load versions of the GIRepository typelib that are not 3.0, and
similarly making gobject-introspection's GIRepository library refuse to
load typelib versions that are not 2.0. A relatively neat way to achieve
that is to make each version behave as if the other one doesn't exist.
Signed-off-by: Simon McVittie <smcv@debian.org>
The title of an interface can be arbitrarily long, considering that
reverse DNS namespaces can be pretty complex. Instead of using the whole
interface name, we can use the name without the prefix.
The library shipped by gobject-introspection.git was
`libgirepository-1.0.so`, but for some reason (accident?), it was
accompanied by `GIRepository-2.0.gir`. That’s been the case for the last
6 years.
In moving libgirepository to glib.git, we’ve bumped the version to
`libgirepository-2.0.so`, and have changed the API.
In order to avoid a collision between the new `GIRepository-2.0.gir` and
the old `GIRepository-2.0.gir`, we can either:
* Rename the basename of the library (confusing).
* Re-version the whole thing to 3.0 (would mean it’s completely out of
sync with the rest of glib.git, and would lead to build system
misery).
* Re-version only the GIR file (a bit confusing, but hopefully less
confusing).
So I’ve done the final option: glib.git now ships
`libgirepository-2.0.so` and `GIRepository-3.0.gir`. This avoids
collisions with what’s shipped by gobject-introspection.git, while
hopefully still making some sense.
We considered using version number 2.1 rather than 3.0, but decided
against it because that makes it look like it’s compatible with version
2.0, which it isn’t.
Note that none of these changes touch the
`${prefix}/lib/girepository-1.0` and `${prefix}/share/gir-1.0`
directories. The version numbers in those refer to the versions of the
GIR and typelib file formats, which have not changed.
Signed-off-by: Philip Withnall <pwithnall@gnome.org>
Helps: #3155
They are now installed to (e.g.)
`${prefix}/share/doc/glib-2.0/{glib,gmodule,gobject,gio}/index.html`.
We might want to drop one level of nesting out of that, but for the
moment I thought I’d keep it in so we can disambiguate by installed
major version.
Signed-off-by: Philip Withnall <pwithnall@gnome.org>
Helps: #3037
Previously, `-Dman=false` was the default, because the generated man
pages were shipped in the distribution tarball already, so the option
actually mostly controlled whether to *re*build them.
The generated pages are no longer shipped in the tarball (and probably
haven’t been since the port to Meson, though I haven’t checked), so it
makes sense to change the default to encourage building the man pages if
the right tooling (`rst2man`) is available.
Signed-off-by: Philip Withnall <pwithnall@gnome.org>
So they are consistent with the way we’re building man pages in other
projects, and because some people are allergic to XML.
This changes the build-time dependencies from `xsltproc` to `rst2man`,
and also takes the opportunity to change the `-Dman` Meson option from a
boolean to a feature (so you should use `-Dman-pages={enabled,disabled}`
now, rather than `-Dman={true,false}`).
Signed-off-by: Philip Withnall <pwithnall@gnome.org>
Helps: #3037
We used to store the search paths into a GSList but this is not
efficient for various reasons, so replace this with an array so that we
can replace return just a GStrv in the public API.
Previously:
- if the object is currently not frozen, we called
g_object_notify_queue_freeze() once. Afterwards dispatch the event
directly. This is probably the common case, and requires one
notify_lock lock.
- if the object is currently frozen, we call
g_object_notify_queue_freeze(), g_object_notify_queue_add().
g_object_notify_queue_thaw().
This required taking the notify_lock three times.
- if the object is currently not frozen and in_init, then we called
g_object_notify_queue_freeze(), g_object_notify_queue_freeze(),
g_object_notify_queue_add(). This also required to take
the lock three times. There is another thaw at the end of
object initialization.
That was because we first call g_object_notify_queue_freeze() to see
whether we are frozen. And depending on that, queue the event (and thaw
again).
Instead, g_object_notify_queue_add() can do the check and queueing in
one step. There is no need to call a freeze() to (conditionally) enqueue
a notification. Now only one lock is taken in all cases.
Also, g_object_notify_queue_freeze() and g_object_notify_queue_thaw()
both call g_datalist_id_get_data() (which also take a bit lock). As the
thaw is no longer necessary, the second lock is also saved.
Before dispatching signals (and calling out to user code), we want to
take a reference and ensure that the object stays alive.
However, a thaw may not decrease the freeze_count to zero, or there may
be no properties to notify. Avoid taking a reference in those cases.
This was done since the beginning (commit e773d7dba6 ('fixed dealing
with collection/lcopy of NULL values.'). But it's not clear, why we
would need to take a reference on the calling object.
Freeze does not emit any signals/callbacks and does not call back to the
user. It just sets up some internal state.
This doesn't require to take a reference. The caller must hold a valid
reference to being with, but if that's given, there is no need to
acquire another reference.
g_atomic_int_get() returns a signed int. While we don't expect this to be ever
negative, a negative value would also indicate a bug. Adjust the check to assert
against negative ref-count too.
Without this when compiling GLib with address sanitizer enabled, we'd
end up failing with this error:
==375535==ASan runtime does not come first in initial library list;
you should either link runtime to your application or manually preload
it with LD_PRELOAD.
Now, given that addressing the fix implies doing more radical changes,
it's just fine here to ignore ASAN to work on tools we use for building
gir files.
Generating gir and typelib files has inter-dependencies that may depend
on other elements.
For example, glib requires gobject and gdump generated files require
gmodule, so we've a cyclic dependency because gmodule requires gobject,
that requires glib.
To prevent this, let's just generate the introspection files at once in
a different meson file so that we don't have to deal with this.
As per this we could even revert commit fa37ab6d0 since gio is now
compiled before the gir files.
GLib gir requires glib-types.h that also includes gobject-visibility.h
that needs to be generated in order to be able to generate the GLib gir,
so explicitly add it to the sources.
Also, in order to link the typelib we need the definitions of the *_get_type()
functions that are defined in the gboxed.c file, so this should be part
of the gir sources or we'll have linking issues.
