Re: Good Use of GOTO
c.k.lester wrote:
>
> > but the following simpler version is at least twice as fast:
>
> As I suspected, there will probably always be a way to optimize code such
> that GOTO is unnecessary /for speed/.
Since euphoria turns loops into goto's in the backend anyways, this may hold
true for the case where we can use a loop.
I think Jeremy had some good examples of cases where one could not always rely
on being able to do this.
>
> We must now consider other "good uses" of GOTO in Euphoria:
>
> 1. porting code from languages that use GOTO
> 2. giving programmer's an extra tool to use if they want
> 3. allowing improved structure (in rare cases)
>
> I think the strongest argument for GOTO is #1, but even that is diminished
> with the integration of PCRE into the core, and the fact that code with GOTO
> has been ported just fine by others in this forum. GOTO can be refactored
> into structured code and still maintain its speed.
PCRE was one example where goto was a make or break deal. It is probably not the
only one.
Rouge can be ported without using goto ok, but as we've already seen that's not
always the case.
>
> #2 is not compelling because the same can be said about pointers, etc.
Euphoria code can use pointers from C code just fine. It even provides the
ability to use function pointers!
>
> Alas, there remains no good reason to include GOTO.
You did not address #3. I do not think it is so rare, since the there is at
least one major project that uses goto extensively, but in a way to make the code
very well structured. Granted, its not written in Euphoria.
|
Not Categorized, Please Help
|
|
