Names immediately on the left side of copula are treated differently
from other names.
In other words, name=. expression is different from (name)=. expression
So I like to emphasize this by ensuring no space immediately to the
left of copula.
Meanwhile, my gut feeling is that for. loops should be slightly (but
usually unmeasurably) faster than equivalent while. loops. The general
guideline would to leave things to the implementation when it provides
the appropriate mechanisms (at least initially).
(And, I hope you enjoy tomorrow - including breakfast.)
Thanks,
--
Raul
On Sun, Jun 25, 2017 at 2:18 PM, 'Mike Day' via Programming
<programm...@jsoftware.com> wrote:
> Thanks.
>
> I haven't heard of paste & match - I like to use Thunderbird to support the
> sharewarers,
> but fight against some of its features. With time & patience, I copy &
> paste into a notepad
> file, and then re-copy and paste into T/bird. I hadn't bothered in my
> recent posts.
> I do (and did in this case) indent and insert whitespace, albeit by hand
> rather than using
> any formatter that J might have to hand.
>
> I see "because copulas are special" but don't quite see why avoiding the
> space helps.
>
> Yes - if I'm really worried about performance, I do similar tweaks on loop
> input. Is a while
> loop better in this case? (I haven't time to check that now.)
>
> Anyway, I've got to practise a bit of music and then go out for the
> evening, so enough
> from me for now.
>
> Breakfast tomorrow is another day,
>
> Mike
>
>
>
>
> On 25/06/2017 18:36, Raul Miller wrote:
>>
>> Not quite, I have a few trivial tweaks that I think make reading it a
>> bit easier.
>>
>> (1) I used "Paste and match style" rather than simply paste, when
>> inserting the code into my message, to eliminate the blank lines that
>> some insane programming teams decided should be inserted when using
>> paste.
>>
>> (2) I indented the code to reflect the explicit structures.
>>
>> (3) I commuted a multiplication to better show similarity with a
>> multiplication on the line above.
>>
>> (4) I ate the whitespace to the left of copulas, because copulas are
>> special.
>>
>> These are, granted, issues of taste, which means there will always be
>> someone with different views. They do not help0 performance nor reduce
>> resource costs in any way. Nevertheless, these reflect my views.
>>
>> And, (5) a nearly invisible performance improvement (eliminate one
>> operation) when setting up the loop.
>>
>> Now... if I could just get my mailer to display messages using a fixed
>> width font...
>>
>> finalrev=: 3 : 0 NB. good name?
>> m=. >./y
>> for_i. 2+i.<:n=. #y do.
>> m=. m >. i * q=. >./ y =. 2 <./\ y
>> if. m >: n * q do. break. end.
>> end.
>> )
>>
>> I am not certain though if my (5) hurts readability, but for code like
>> this you sort of need to go in and see the data to really understand
>> it. So, here's an instrumented version - I would recommend only trying
>> this on small arguments (and the displayed result is meant to be used
>> to help understand the uninstrumented version):
>>
>> show=: 1 {:: 1!:2&2@(] :(,~))
>> showfinalrev=: 3 : 0
>> show 'y';y
>> show 'm';m=. >./y
>> for_i. 2+i.<: show 'n';n=. #y do.
>> ('q';q;'y';y) show 'm';m=. m >. i * q=. >./ y =. 2 <./\ y
>> if. show 'break';m >: n * q do. break. end.
>> end.
>> )
>>
>> That said, thank you for the excellent treatment of this problem. You
>> took this a lot further than I had.
>>
>> Thanks,
>>
>
>
> ---
> This email has been checked for viruses by Avast antivirus software.
> https://www.avast.com/antivirus
>
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
