[m-dev.] for review: switches via binary search

[Fergus Henderson]

On 30-Dec-1997, Zoltan Somogyi <zs at cs.mu.oz.au> wrote:
> 
> Fergus, can you please review this? I'd ask Tom, but he won't be around
> for a while.
> 
> compiler/tag_switch.m:
> 	Add a new way of generating code for switches, binary search chains.

I don't think "chain" is the right term to use to describe it.

How about
 	Add a new way of generating code for switches, using binary search.

options.m:
> +	io__write_string("\t--binary-switch-size <n>\n"),
> +	io__write_string("\t\tThe number of alternatives in a binary search chain switch\n"),
> +	io__write_string("\t\tmust be at least this number (default: 4).\n"),

I suggest deleting the word "chain" here.
(Ditto for doc/user_guide.texi.)

tag_switch.m:
> +	% 4. binary search chains have the form

How about
	% 4. binary searches have the form

> +	% For binary search chains, we want the case of the most frequently

Ditto.

> +:- type switch_method	--->	try_me_else_chain
> +			;	try_chain
> +			;	jump_table
> +			;	binary_chain.

I suggest s/binary_chain/binary_search/g

>  tag_switch__generate(Cases, Var, CodeModel, CanFail, StoreMap, EndLabel, Code)
...
> -	( { PtagsUsed > DenseSwitchSize } ->
> +	( { PtagsUsed >= DenseSwitchSize } ->
...
> -	; { PtagsUsed > TrySwitchSize } ->
> +	; { PtagsUsed >= TrySwitchSize } ->

I think those changes weren't mentioned in the log message.

I haven't really looked in any detail at the new slabs of code,
but if it works, then apart from the things mentioned above I'm
happy with it.

-- 
Fergus Henderson <fjh at cs.mu.oz.au>   |  "I have always known that the pursuit
WWW: <http://www.cs.mu.oz.au/~fjh>   |  of excellence is a lethal habit"
PGP: finger fjh at 128.250.37.3         |     -- the last words of T. S. Garp.