[m-rev.] for review: fix some failing decl debugger test cases

[Ian MacLarty]

On Tue, Jan 18, 2005 at 05:01:53PM +1100, Julien Fischer wrote:
> 
> On Sun, 16 Jan 2005, Ian MacLarty wrote:
> 
> > Estimated hours taken: 12
> > Branches: main
> >
> > Disable higher order specialization for the mdb.declarative_edt module
> > since mdb.declarative_edt.revise_root/3 exposes a bug in this optimization.
> >
> I take it that this is the reason all of those test cases for the
> declartive debugger are failing on earth and jupiter?
> 

Yup.

> 
> > Allow the declarative debugger to continue to be debugged after a new portion
> > of the annotated trace has been generated.
> >
> > browser/Mercury.options
> > 	Disable higher order specialization for the mdb.declarative_edt module.
> >
> > trace/mercury_trace_declarative.c
> > trace/mercury_trace_declarative.h
> > 	Add a global variable MR_trace_decl_in_dd_dd_mode to indicate when
> > 	the declarative debugger was invoked with the dd_dd command.  A
> > 	seperate variable must be used to trace_mode, since trace_mode
> s/seperate/separate/
> 
> Should that be "...must be used for trace_mode, since ..."?
> 

No.  I want to say that trace_node can't be used to remember whether the
declarative debugger was invoked with the dd_dd command.

> > 	is changed to MR_TRACE_INTERACTIVE when the declarative debugger
> > 	is started, so it's not possible to tell that it was invoked with the
> > 	dd_dd command when diagnosis is restarted after a new portion of the
> > 	annotated trace has been generated.
> >
> > trace/mercury_trace_internal.c
> > 	Set MR_trace_decl_in_dd_dd_mode appropriately depending on whether
> > 	`dd' or `dd_dd' was used to invoke the declarative debugger.
> >
> > 	When the dd_dd command is used set the trace_mode to
> I'd a comma between "used" and "set" there.
> 

Done.

> > 	MR_TRACE_DECL_DEBUG instead of MR_TRACE_DECL_DEBUG_DEBUG since the
> > 	value of MR_trace_decl_in_dd_dd_mode is now used instead to determine
> > 	if the dd_dd command was given.
> >
> > trace/mercury_trace_internal.h
> > 	Remove the MR_TRACE_DECL_DEBUG_DEBUG trace mode option since it is
> > 	no longer used.
> >
> > Index: browser/Mercury.options
> > ===================================================================
> > RCS file: /home/mercury1/repository/mercury/browser/Mercury.options,v
> > retrieving revision 1.4
> > diff -u -r1.4 Mercury.options
> > --- browser/Mercury.options	19 Oct 2004 06:01:36 -0000	1.4
> > +++ browser/Mercury.options	14 Jan 2005 04:33:22 -0000
> > @@ -9,5 +9,10 @@
> >  MCFLAGS-mer_browser = --no-warn-nothing-exported
> >  MCFLAGS-mer_mdbcomp = --no-warn-nothing-exported
> >
> > +# XXX Disable higher order specialization for declarative_edt since this
> > +# optimization has a suspected bug that mdb.declarative_edt.revise_root/3
> > +# exposes.
> It is a bug, not just a suspected one.
> 

Yes, but I'm not 100% sure that the bug is in higher order specialisation (even
though when I disable this optimization it works).

> > +MCFLAGS-mdb.declarative_edt = --no-optimize-higher-order
> > +
> >  EXTRA_LIBRARIES-libmer_browser.so = mer_mdbcomp
> >  EXTRA_LIBRARIES-libmer_browser.dylib = mer_mdbcomp
> 
> > Index: trace/mercury_trace_declarative.h
> > ===================================================================
> > RCS file: /home/mercury1/repository/mercury/trace/mercury_trace_declarative.h,v
> > retrieving revision 1.21
> > diff -u -r1.21 mercury_trace_declarative.h
> > --- trace/mercury_trace_declarative.h	9 Jan 2005 01:14:09 -0000	1.21
> > +++ trace/mercury_trace_declarative.h	14 Jan 2005 04:44:37 -0000
> > @@ -127,4 +127,13 @@
> >
> >  extern	MR_Integer	MR_edt_depth_step_size;
> >
> > +/*
> > +** The following variable indicates whether the declarative debugger was
> > +** invoked using the dd_dd command and is needed so that the declarative
> Make this comment into two separate sentences.  e.g.
> 
> 	It is needed so that the declarative ...
> 

Done.

Cheers,
Ian.
--------------------------------------------------------------------------
mercury-reviews mailing list
post:  mercury-reviews at cs.mu.oz.au
administrative address: owner-mercury-reviews at cs.mu.oz.au
unsubscribe: Address: mercury-reviews-request at cs.mu.oz.au Message: unsubscribe
subscribe:   Address: mercury-reviews-request at cs.mu.oz.au Message: subscribe
--------------------------------------------------------------------------