1=====
2Goals
3=====
4
5The introspection project has two major goals, and a variety of more minor
6ones.
7
8
9Two level applications - C and <your favorite runtime>
10------------------------------------------------------
11
12It makes sense to build many kinds of applications using (at least) two
13different levels and languages — one for the low level elements, interfacing
14with the OS and/or the hardware; and one for the high level application logic.
15C is good for graphics, multimedia, and lower level systems work. However,
16writing complex software in C is difficult and error-prone. A managed runtime
17such as `JavaScript <https://wiki.gnome.org/JavaScript>`__, Python, Perl,
18Java, Lua, .NET, Scheme etc. makes a lot of sense for non-fast-path
19application logic such as configuration, layout, dialogs, etc.
20
21
22.. note::
23
24  To achieve this goal you need to write your code using GObject convention.
25  For more information about that, see the `GObject tutorial
26  <https://developer-old.gnome.org/gobject/stable/pt02.html>`__
27
28Thus, one of the major goals of the GObject introspection project is to be a
29convenient bridge between these two worlds, and allow you to choose the right
30tool for the job, rather than being limited inside one or the other. With the
31introspection project, you can write for example a ClutterActor or GtkWidget
32subclass in C, and then without any additional work use that class inside the
33high level language of your choice.
34
35
36Sharing binding infrastructure work, and making the platform even more binding-friendly
37---------------------------------------------------------------------------------------
38
39Historically in GNOME, the core platform has been relatively binding-friendly,
40but there are several details not captured in the C+GObject layer that
41bindings have needed. For example, reference counting semantics and the item
42type inside GList's. Up until now various language bindings such as Python,
43Mono, java-gnome etc. had duplicated copies of hand-maintained metadata, and
44this led to a situation where bindings tended to lag behind until these manual
45fixups were done, or were simply wrong, and your application would crash when
46calling a more obscure function.
47
48The introspection project solves this by putting all of the metadata inside
49the GObject library itself, using annotations in the comments. This will lead
50to less duplicate work from binding authors, and a more reliable experience
51for binding consumers.
52
53Additionally, because the introspection build process will occur inside the
54GObject libraries themselves, a goal is to encourage GObject authors to
55consider shaping their APIs to be more binding friendly from the start, rather
56than as an afterthought.
57
58
59Additional goals and uses
60-------------------------
61
62* API verification - Sometimes the API of a library in our stack changes by
63  accident. Usually by a less experienced developer making a change without
64  realizing it will break applications. Introspecting the available API in
65  each release of the library and comparing it to the last one makes it easy
66  to see what changed
67* Documentation tools - The tools written inside of the GObjectIntrospection
68  can easily be reused to improve that problem. Essentially; replacing
69  gtk-doc. We want to document what we export so it makes sense to glue this
70  together with API verification mentioned above
71* UI Designer infrastructure
72* Serialization/RPC/DBus
73