Re: [Bug-apl] Unique far too slow

[Ala'a Mohammad]

Thanks to all!

Now unique went from 2 min and 36 sec down to 10 sec, in the
spellcheck script.

Ala'a


On Sat, Sep 17, 2016 at 4:43 PM, Juergen Sauermann
<address@hidden> wrote:
> Hi Xtian,
>
> thanks, fixed in SVN 796.
>
> /// Jürgen
>
>
> On 09/17/2016 05:12 AM, Christian Robert wrote:
>
>
>       time "z ← downcase ¨ words readfile 'big.txt'"
> 7.859168565
>
>       time "s ← sortvs z"
> 4.423386895
>
>       ⍝ well, read took 8 seconds (over a million words)
>       ⍝ sorting the result took 5 seconds
>
>       time "U ← ∪ z"
> 101.1162267
>
>       ⍝ why does the "∪" operator took so long ? eg: 101  seconds.
>
>
> pretty sure this can be improved by a factor 2 to 5 times shorter.
>
>
>
> my definitions:
>
>   downcase←{ ⎕UCS (32×(⍵≥65)∧⍵≤90)+⍵←⎕UCS ⍵}
>   readfile←{⎕fio[26] ⍵}
>   sortvs←{⍵[⎕av⍋⊃⍵]}
>   words←{(⍵ ∊ ,⎕ucs (65 97)∘.+⎕io-⍨⍳26) ⊂ ⍵}
>
> big.txt is the one used in several reports last 2 weeks.
>
>
> Xtian.
>
>