Mark Wooding [Sat, 15 Aug 2015 15:25:24 +0000 (16:25 +0100)]
src/class-output.lisp: Insert the correct class pointers.
I must have been confused, because vtable class pointers ended up
pointing to the metaclass rather than the object class. Fix this.
Mark Wooding [Sat, 15 Aug 2015 15:24:29 +0000 (16:24 +0100)]
src/class-output.lisp: Output effective methods directly from the class.
Earlier there was a bug where effective methods and method entries were
emitted twice, because there were two paths to them: one directly from
the class, and one via method entry points. To resolve this, I removed
the path directly from the class.
This turns out to have been the wrong choice, because method entries are
found once for each chain, so there are still duplicates in classes with
more than one chain carrying methods. So don't iterate from method
entries to effective methods, and reinstate the direct path from the
top-level class.
Mark Wooding [Sat, 15 Aug 2015 15:19:47 +0000 (16:19 +0100)]
src/parser/scanner-proto.lisp: Reorder declarations.
The `token-scanner' and `token-scanner-place' objects refer to each
other recursively through their slot types. SBCL's loader objects to
this if `token-scanner-place' -- a structure type -- is first. Reorder
them so that `token-scanner' -- a CLOS class -- is first, because that
seems to work better.
Mark Wooding [Sat, 15 Aug 2015 15:16:25 +0000 (16:16 +0100)]
src/class-output.lisp: Fix docstring grammar.
Mark Wooding [Sat, 15 Aug 2015 15:15:58 +0000 (16:15 +0100)]
src/builtin.lisp: Delete unused variable binding.
Mark Wooding [Sat, 15 Aug 2015 15:15:16 +0000 (16:15 +0100)]
src/foo.sod: Split `Second' off into its own separate chain.
Mark Wooding [Sat, 15 Aug 2015 15:14:32 +0000 (16:14 +0100)]
lib/sod.h: New macro `SOD_INSTBASE' to find the allocated base address.
This turns out to be useful, e.g., for freeing dynamically allocated
objects. A destructor method might usefully say something like
free(SOD_INSTBASE(me));
after actually releasing any other resources.
Mark Wooding [Sat, 15 Aug 2015 02:28:59 +0000 (03:28 +0100)]
An actual running implementation, which makes code that compiles.
I've not tested it very hard yet. There are probably lots of small
bugs, and maybe some big ones. There are certainly rough edges in the
error handling, particularly in the class-body parser which will get
cleaned up later.
Mark Wooding [Wed, 24 Jul 2013 21:54:12 +0000 (22:54 +0100)]
More WIP.
Mark Wooding [Tue, 23 Jul 2013 23:49:14 +0000 (00:49 +0100)]
Today's wip.
Mark Wooding [Wed, 17 Jul 2013 20:03:27 +0000 (21:03 +0100)]
Refactoring more or less complete. Maybe I should test it.
Mark Wooding [Mon, 15 Jul 2013 00:54:09 +0000 (01:54 +0100)]
Daily work in progress.
Mark Wooding [Sun, 14 Jul 2013 15:09:22 +0000 (16:09 +0100)]
Work in progress. Mostly bug fixing.
Mark Wooding [Fri, 12 Jul 2013 01:39:37 +0000 (02:39 +0100)]
Fix formatting badness.
Mark Wooding [Fri, 12 Jul 2013 01:25:39 +0000 (02:25 +0100)]
Remove apparently dead files.
Mark Wooding [Fri, 12 Jul 2013 01:24:21 +0000 (02:24 +0100)]
Don't write Lisp symbol names in uppercase: use `...' instead.
Otherwise it's too confusing to distinguish them from metavariables.
Mark Wooding [Fri, 12 Jul 2013 00:37:09 +0000 (01:37 +0100)]
Change naming convention around.
Put `test', `proto', `impl', `parse', `output' on the end, not the
start. This keeps pieces of the same chunk of functionality together in
listings.
Mark Wooding [Fri, 12 Jul 2013 00:22:58 +0000 (01:22 +0100)]
Work in progress, recovered from old crybaby.
Mark Wooding [Fri, 2 Jul 2010 09:11:35 +0000 (10:11 +0100)]
Massive reorganization in progress.
The code is a complete disaster area right now.
Mark Wooding [Tue, 20 Oct 2009 00:42:17 +0000 (01:42 +0100)]
It lives!
The module parser is more-or-less done. Output is more-or-less done.
Outstanding. Now to remove the bugs...
Mark Wooding [Sat, 17 Oct 2009 23:41:01 +0000 (00:41 +0100)]
Add .gitattributes file.
Mark Wooding [Sat, 17 Oct 2009 23:05:32 +0000 (00:05 +0100)]
Another day, another commit.
Main highlight is that we now have output of class instances and direct
methods. Yet to come are effective methods and entries, and vtables.
Today's work has been surprisingly focussed. Incidental changes are
few; the most significant is probably trimming trailing whitespace from
the last line of a fragment, which eliminates some ugly blank lines in
the output.
Mark Wooding [Sat, 17 Oct 2009 00:10:34 +0000 (01:10 +0100)]
Lots more has happened.
* all reserved words are now banished
* an ilayout now consists of, for each chain, a union of the ichains
for each class on the chain -- makes referring to the bit which
represents a superclass instance trivial; not yet hacked the
effective method codegen to cope
* unnecessary ichain and vt structures (ones which are copies of a
superclass's) are suppressed -- we use the original directly
(must document the new chain-tail tracking stuff); only one new
ichain and vtable structure per class (though we still need all-new
vtmsgs for each superclass).
* header file output is complete and functional
Mark Wooding [Thu, 15 Oct 2009 09:24:32 +0000 (10:24 +0100)]
More work. Highlights:
* start using new module import protocol
* move bootstrapping to separate file
* eliminated most of the non-C reserved words
Maybe I'll eliminate the C reserved words too. It'll make the grammar
easier to extend.
Mark Wooding [Wed, 14 Oct 2009 22:39:31 +0000 (23:39 +0100)]
Another day.
* Wrote (but haven't compiled!) C runtime code.
* Fettled runtime header.
* Reformatted class definitions. They don't deserve that much space.
Revamped module protocol to come.
Mark Wooding [Wed, 14 Oct 2009 00:17:21 +0000 (01:17 +0100)]
Very ragged work-in-progress.
Most parts are in place. Much rearrangement is needed.
Mark Wooding [Wed, 14 Oct 2009 00:09:19 +0000 (01:09 +0100)]
Early work-in-progress.
Much needs to be done.