removing unused variables?

removing unused variables?

[Andy Wingo]

Hello,

Unused variables are compiled away to only their side effects. Yet
sometimes they are good for documentation. As in this case:

------------------------- module/language/assembly.scm -------------------------
index 683da6c..95f8a2d 100644
@@ -49,7 +49,7 @@
      (+ 1 *len-len* (string-length str)))
     ((load-array ,bv)
      (+ 1 *len-len* (bytevector-length bv)))
-    ((load-program ,nargs ,nrest ,nlocs ,labels ,len ,meta . ,code)
+    ((load-program _ _ _ _ ,len ,meta . _)
      (+ 1 *program-header-len* len (if meta (1- (byte-length meta)) 0)))
     ((,inst . _) (guard (>= (instruction-length inst) 0))
      (+ 1 (instruction-length inst)))


This is a bad change IMO. We should not contort our code to please some
mechanical idea of "good style".

Andy
-- 
http://wingolog.org/

Re: removing unused variables?

[Ludovic Courtès]

Hi,

Andy Wingo <wingo@pobox.com> writes:

> ------------------------- module/language/assembly.scm -------------------------
> index 683da6c..95f8a2d 100644
> @@ -49,7 +49,7 @@
>       (+ 1 *len-len* (string-length str)))
>      ((load-array ,bv)
>       (+ 1 *len-len* (bytevector-length bv)))
> -    ((load-program ,nargs ,nrest ,nlocs ,labels ,len ,meta . ,code)
> +    ((load-program _ _ _ _ ,len ,meta . _)
>       (+ 1 *program-header-len* len (if meta (1- (byte-length meta)) 0)))
>      ((,inst . _) (guard (>= (instruction-length inst) 0))
>       (+ 1 (instruction-length inst)))
>
>
> This is a bad change IMO. We should not contort our code to please some
> mechanical idea of "good style".

Oops, I actually agree with this one.  Probably I should not have
committed it before discussion, or at least in a separate, easily
revertable commit.  I apologize for not doing so.

I can revert the offending parts, which are all ‘pmatch’ invocations.
Would that be OK with you?

That said, in many cases unused variables are a sign of sloppiness IMO,
which was the reason I looked into it.  However, having unused variables
“for style” and unused variables introduced by macros makes it harder to
identify “really unused” variables.

Thanks,
Ludo’.

Re: removing unused variables?

[Andy Wingo]

Hi Ludovic,

On Mon 21 Sep 2009 11:08, ludo@gnu.org (Ludovic Courtès) writes:

> Hi,
>
> Andy Wingo <wingo@pobox.com> writes:
>
>> ------------------------- module/language/assembly.scm -------------------------
>> index 683da6c..95f8a2d 100644
>> @@ -49,7 +49,7 @@
>>       (+ 1 *len-len* (string-length str)))
>>      ((load-array ,bv)
>>       (+ 1 *len-len* (bytevector-length bv)))
>> -    ((load-program ,nargs ,nrest ,nlocs ,labels ,len ,meta . ,code)
>> +    ((load-program _ _ _ _ ,len ,meta . _)
>>       (+ 1 *program-header-len* len (if meta (1- (byte-length meta)) 0)))
>>      ((,inst . _) (guard (>= (instruction-length inst) 0))
>>       (+ 1 (instruction-length inst)))
>>
>>
>> This is a bad change IMO. We should not contort our code to please some
>> mechanical idea of "good style".
>
> Oops, I actually agree with this one.  Probably I should not have
> committed it before discussion, or at least in a separate, easily
> revertable commit.  I apologize for not doing so.
>
> I can revert the offending parts, which are all ‘pmatch’ invocations.
> Would that be OK with you?

That would be great.

> That said, in many cases unused variables are a sign of sloppiness IMO,
> which was the reason I looked into it.  However, having unused variables
> “for style” and unused variables introduced by macros makes it harder to
> identify “really unused” variables.

Yes, agreed with all of this.

Thanks! (And apologies if I sounded grumpy. Perhaps I had mailed before
coffee ;-)

Andy
-- 
http://wingolog.org/

Re: removing unused variables?

[Ludovic Courtès]

Hi Andy,

Andy Wingo <wingo@pobox.com> writes:

> On Mon 21 Sep 2009 11:08, ludo@gnu.org (Ludovic Courtès) writes:

>> I can revert the offending parts, which are all ‘pmatch’ invocations.
>> Would that be OK with you?
>
> That would be great.

Done.

> Thanks! (And apologies if I sounded grumpy. Perhaps I had mailed before
> coffee ;-)

No problem.  ;-)

Ludo’.