From: Craig Chambers (chambers@cs.washington.edu)
Date: Mon Oct 07 2002 - 18:04:29 PDT
I'm publishing some more miscellaneous improvements to the whole
Vortex/Whirlwind system.
*) I extended the SETUP scripts to set up Whirlwind's required
libraries and binaries, too.
*) I futzed with the code for parsing & tokenizing C primitives, to
account better for the potential for braces and quoting to be
embedded in comments or strings. I fixed a problem in Whirlwind's
change testing code that related to this possibility. (E.g., if you
used an apostrophe in a C++ comment in a primitive in Wil code, then
future changes to that function could lead to parse errors due to a
misformed Wil character constant.) I rewrote most of the C
primitives in the Vortex codebase (all but those in motif.cecil) to
use the prim c_++ { ... } syntax instead of the prim c_++: "..."
syntax; the former doesn't require backslashing of all " and \ in the
prim body, and the latter now works fine even if the C++ primitive
uses ' and { in comments and strings. (I think Vortex could already
work fine for this (the previous standard Vortex executable can still
parse the new code base), but Whirlwind couldn't until I made the
changes.)
*) I added a little more code to the class hierarchy analysis code.
It still doesn't do anything interesting, but more of the framework
is there.
*) I added some prohibitions in the rep-checking pass of Whirlwind to
rule out class declarations in nested scopes; the compiler only
allows class declarations at the top level. (Nested function
declarations are allowed, as long as they do not reference lexically
enclosing non-global variables. Nested gf declarations can be
compiled properly, but they cannot be invoked at run-time, because
they require an unimplemented generic dispatching strategy. Nested
method declarations are allowed, but they should only be applied to
gfs in the same scope; I don't think this last restriction is
enforced yet.) I added some test code for nested declarations to the
Wil regression test suite.
*) I made some other small improvements to the regression test suite.
*) I added some refined rep-checking code for int size conversion
operators.
-- Craig
_______________________________________________
Cecil mailing list
Cecil@cs.washington.edu
http://majordomo.cs.washington.edu/mailman/listinfo/cecil
This archive was generated by hypermail 2.1.5 : Mon Oct 07 2002 - 18:04:34 PDT