[m-dev.] status of 0.12 branch

[Julien Fischer]

The following is a list of things are broken on the
0.12 branch.  Most of these apply to the main branch as well.

The following are broken:

* the clp(r) interface
* extras/trailed_update

	Both of these things use the old version of
	solver types and haven't been updated.

* the deep profiling tool

* the compiler allows non-abstract typeclass instances
  in modules interfaces - even when the typeclass itself is
  abstract.  For example, it compiles the following without
  complaint:

	:- module instance_in_interface.

	:- interface.

	:- import_module int.

	:- typeclass foo(T).

	:- instance foo(int) where [
		id(X) = X
	].

	:- implementation.

	:- typeclass foo(T) where [
		func id(T) = T
	].

	:- end_module instance_in_interface.

The following test case are failing:

* tests/valid/unused_args_2.m is causing the following assertion
  failure in the compiler:

	Uncaught Mercury exception:
	Software Error: map__lookup: key not found
		Key Type: term.var(parse_tree.prog_data.prog_var_type)
		Key Value: var(2)
		Value Type: ll_backend.var_locn.var_state
	Stack dump not available in this grade.

  This bug can be reproduced by compiling with:

	 mmc --grade asm_fast.gc -C -O0 --optimize-saved-vars-const unused_args_test2

* tests/debugger/declarative/catch is failing in the debug grades because
  there is no tracing code in builtin_catch.  The long term plan here
  is to modify builtin_catch.  In the meantime I think we should just have
  the debugger print an error message that it cannot perform retries
  over caught exceptions.

* tests/debugger/field_names is failing in the .decldebug grades.
  (Ian explained this one to me but I've forgotten what the explanation
   for this failing is).

* tests/debugger/solver_test fails when compiled with --intermodule-optimization.
  This seems to be because the `procedures' command in the debugger is
  printing out, not just the procedures from the solver_test module, but
  also procedures from any imported/opt_imported modules(?).

* tabling/memo_non is failing in the deep profiling grades.
  I asked Zoltan about this some time ago and he didn't think it was
  a problem.  Now that we using tabling in the library is it a problem?
  If not, should we just disable this test for the deep profiling grades?

--------------------------------------------------------------------------
mercury-developers mailing list
Post messages to:       mercury-developers at cs.mu.oz.au
Administrative Queries: owner-mercury-developers at cs.mu.oz.au
Subscriptions:          mercury-developers-request at cs.mu.oz.au
--------------------------------------------------------------------------