Xaml Compiler: SoC Closure status report

[Iain McCoy]

Hello folks.

It is now September the Second, 2005, local time, so I'm late, but it's
the antipodes, so there's lots of time relative to those Other Timezones
that aren't holes and I suppose therefore I'm actually running early.

We weren't asked for a final report, but I figure I may as well, so here
it is.

Last time was penultimate, so this is, I suppose, ultimate. How extreme.

I've just made a few changes (mostly to make the code more
comprehensible, at least in theory) and I now think the code is
satisfactory for SoC purposes. I consider the xaml stuff in revision
49270 of the mono svn repos to be the "End Of SoC" revision, I hope
that's vaguely reasonable.

There's a few things that I still plan to do, and I do want to push it
to try to get some use.

Here's what's done:

1. A complete compiler from xaml to code, no known bugs
2. A near-complete interpreter from xaml to an object instance - there's
one feature I only found on monday, which is supported in the compiler
but has a bug in the interpreter. I plan to find and fix that bug
soon-ish, but since the feature is useful mainly for supporting the
styling system I don't see it's lack hurting many people for the time
being. That's the only bug I know of.
3. A partial implementation of System.Windows.DependencyObject. This is
the foundation of the new event/property/styling/animation system, and
I've written much of the property bits.
4. The bits of the System.Windows.Serialization namespace that I could
work out reasonable uses for.

So, that's pretty much what I said I'd do in the proposal/acceptance
criteria.

It's mostly in mcs/class/PresentationFramework (the System.Windows stuff
is in mcs/class/WindowsBase), but the compiler and examples are in
mcs/tools/xamlc. Documentation, such as it is, is in
mcs/tools/xamlc/README. You could learn a lot from the tests in
mcs/class/PresentationFramework/Test, more from Parser.cs and
ParserToCode.cs than from XamlParser.cs (that last one is mostly testing
fairly ugly internals).

I think it's at a stage where those directories could reasonably be
built and installed by default.

Things I still plan to do, despite them being outside the scope of the
project:
1. Get users
2. implement the styling system in System.Windows (I didn't do this
during SoC because it looks like requiring a bunch of other
DependencyObject stuff that is almost entirely irrelevant to xaml)
3. act as maintainer-ish person for this code, as required.

Uh, I think that's about all.

Oh, quick brag: I rode my unicycle (29" wheel, big apple 2.35" tire) the
20km from my house to university this morning. Took about 2 hours, with
almost no stops (for those that didn't know this already: unicycles
don't freewheel). Go me. I'm going to work towards doing the ride in 1
hour and 30 minutes, don't know if I'm capable of that sort of time or
not.

Well, I have about 3 slightly-neglected university deadlines approaching
- two tests and an assignment - so for a time I will refocus myself on
such boring things, and once they ease off I can hopefully get some more
work done on this stuff.

Hope everyone else is satisfied with their projects (and has had fun),
-Iain

[Iain McCoy]

On Fri, 2005-09-02 at 00:57 +1000, Iain McCoy wrote:
> I've just made a few changes (mostly to make the code more
> comprehensible, at least in theory) and I now think the code is
> satisfactory for SoC purposes. I consider the xaml stuff in revision
> 49270 of the mono svn repos to be the "End Of SoC" revision, I hope
> that's vaguely reasonable.

And from the spoke-too-soon department, I just realized that I'd
forgotten to add a file to svn. Oops. There were two other forgotten
files as well; they were all boring, but they're in there now. The
revision with those added is r49275; if the stuff doesn't build, there's
probably some other forgotten file. I don't think there is, but I'm not
certain. Let me know if that's the case.

See you later folks,
-Iain