Closed GoogleCodeExporter closed 9 years ago
The trouble this introduces is that the assert statement needs the
AssertionError
exception and as exceptions become classes (rather than types), the VM will
require
classes in order to support the system tests. In short, this makes
HAVE_CLASSES a
requirement for core functionality. Some smaller-RAM targets will suffer.
Original comment by dwhall...@gmail.com
on 18 Jan 2010 at 4:33
The trouble is remedied by the existing HAVE_ASSERT definition which could be
disabled
for smaller targets alleviating the need for classes; however, system tests
won't be
able to run on such target platforms.
Original comment by dwhall...@gmail.com
on 18 Jan 2010 at 3:34
Removed 3 places where OBJ_TYPE_EXN was used for managing the EXN type's
attributes.
Edited RAISE_VARARGS bytecode to use the new class_isSubclass() function to fix
the bug
mentioned above. Made various fixups where TYPE_EXN was used. Had to create a
new
global "Exception" string so that builtin could be loaded from within interp().
Now
that HAVE_ASSERT requires HAVE_CLASS, it is relocated within all
platform/*/pmfeatures.h. Updated docs/src/AssertStatement.txt (docs build).
Original comment by dwhall...@gmail.com
on 18 Jan 2010 at 10:38
Tests pass. Builds for avr, arm7 and mbed. ipm works.
Mainlined directly, see r410
Original comment by dwhall...@gmail.com
on 18 Jan 2010 at 10:46
Changing Milestone-r10 to Milestone-r09
Original comment by dwhall...@gmail.com
on 7 Mar 2010 at 6:26
Changed r09 to R09
Original comment by dwhall...@gmail.com
on 12 Mar 2010 at 8:22
Original comment by dwhall...@gmail.com
on 13 Mar 2010 at 1:30
r517
Post mainline checkin. Added system tests that should be enabled after
exceptions can be handled.
Original comment by dwhall...@gmail.com
on 9 Jul 2010 at 8:29
Original issue reported on code.google.com by
dwhall...@gmail.com
on 25 Sep 2009 at 2:47