[m-rev.] For review: Modify deep profiling tools to use new datastructures.

[Paul Bone]

On Mon, Jul 28, 2008 at 02:07:13PM +1000, Zoltan Somogyi wrote:
> On 25-Jul-2008, Paul Bone <pbone at csse.unimelb.edu.au> wrote:
> > I'm preparing a seperate patch to introduce sv_append/3 to
> > library/string.m.  Then here I an use sv_append("<br>\n", !HTML)
> 
> I don't think that is very intuitive.
> 
> Simply using 
> 
> 	!:HTML = !.HTML ++ "..."
> 
> would do here.

Fixed.

> > It hasn't been a problem yet.  This module suffers from the general
> > problem that S = S0 ++ S1 ++ .. ++ Sn, runs in time O(N*length(S0) +
> > (N-1)*length(S1) ...) which is O(N^2).
> > 
> > I'd like to leave the code above alone until we're ready to fix the
> > general problem (which is not very high compared to other things we'd
> > like to improve).
> 
> The right fix is probably to change the representation of the HTML from
> a simple string to a cord of strings, which is then converted to a single
> string just before being printed out. I can do that after you check in this
> diff. Most of the code won't have to be changed, since ++ is also the operator
> for concatenating cords.

I would prefer that the string module can do this transparently, but
this is a fair bit more work which is why I didn't bother optimising
this at all.

I agree that using cords is the right approach here.

Otherwise I'll check this code in and we can decide if/how to optimise
it later.

> > Now reads:
> > 
> >     The representation of a procedure in the report structure, including
> >     information about it's location in mercury source code.
> 
> s/it's/its/

Thanks.

I've committed this change.


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mercurylang.org/archives/reviews/attachments/20080729/72b18a29/attachment.sig>