[m-rev.] For review: Annotate the HLDS with regions

[Julien Fischer]

On Fri, 29 Jun 2007, Quan Phan wrote:

>>> +
>>> +	% Instruction is of the form: "Tx: xxxxxx R..."
>>> +	% The first x is the number of the rule, which is not important here.
>>> +	% The next 6 x is either "remove" or "create", the last part is the
>>> +	% region name. This region name will be subjected to renaming due to
>>> +	% if-then-else and region resurrection.
>>> +	% This predicate turns such an instruction into a call to a suitable
>>> +	% region builtin.
>>> +	%
>>
>> Instructions should *not* be represented as strings.  You're using
>> a strongly-typed language, so use types!  e.g.
>>
>>  	:- type region_instruction
>>  		--->	region_instruction(
>>  				ri_rule_num :: int,
>>  				ri_instr_kindg :: region_op,
>>  				ri_name :: ???
>>  			).
>>
>>  	:- type region_op
>>  		--->	region_create
>>  		;	region_remove.
>>
>> or something along those lines.  That would also mean that a lot if
>> the if-then-elses that occur in the following code could be turned into
>> switches.
>>
>
> I agree to use a type for region instructions (we may have more of them in the
> future). What I want to do is to postpone this change a bit after we have a
> runnable system. So I put XXX comment at this place.
> Is that acceptable for now?

Yes.

Julien.
--------------------------------------------------------------------------
mercury-reviews mailing list
Post messages to:       mercury-reviews at csse.unimelb.edu.au
Administrative Queries: owner-mercury-reviews at csse.unimelb.edu.au
Subscriptions:          mercury-reviews-request at csse.unimelb.edu.au
--------------------------------------------------------------------------