Hacker News Reader

Comment by jnbiche | Hacker News Reader

Forget Perl, this is serious write-only software engineering.

I mean, I'm a language polyglot, and I think these languages are probably great for multidimensional array processing. But for general purpose programming, not so much.

But cool hack.

View on HN

Don't dismiss things you don't understand: I think array programming is extremely useful for parsing, and I also think that with some untraining and some practice you would find it easy to read and write.

I have a new programmer- strong maths background, but no programming experience prior to K which she started about two weeks ago.

Today, she's working on taking strings that look like this:

    y:"http://host/path/to.a.swf,http://host/path/to/b.swf"

and finding out if the url-fragments are "claimed" by any any other user in a database, AND if the current user claims any url-fragments. So she has a database like this:

    d:`f`u!(("adap.tv";"adap.tv";"btrll.com";"btrll.com";"cbs.com";"cbsinter…
    l:?d`f

and in about thirty minutes of puzzling came up with this:

    |/#:'y ss/:l[&~|/d.f[&x=d.u]~/:\:l]

Since you don't know K I'll walk you through it:

    d.f[&x=d.u]~/:\:l

This returns a list of bitmaps, such as:

    (10000000000000000b;00000100000000000b)

What it's doing is taking all of the urls and indexing it by the where (&) address of each x (parameter to function: user id) is equal to the array of all of the users. That is, x=d.u returned: 01000001000000000000000000b and &x=d.u returned 1 7 (because the second and eighth element of that bitmap are ones). d.f[1 7] returns the second and eighth fragment.

Then, she &~|/ which ors- the bitmaps (|/) together, finds their complement (~), and takes the where of (&) them, so in this case 0 5. l[0 5] then returns the strings in the master list.

Note this is the list of all of the url fragments owned by other users:

    l[&~|/d.f[&x=d.u]~/:\:l]

In PHP you might try to write:

    $mine = array();
    $theirs = array();
    foreach($db as $row) {
      if($row->u === $x) $mine[$row->f]=0;
      else $thiers[$row->f]=0;
    }
    $list = array_diff(array_keys($thiers),array_keys($mine));

but there's a lot of syntax here, and a number of PHP function calls that a beginner might not be aware of. She's just using this:

    http://www.timestored.com/kdb-guides/files/kdb-cheat-sheet.pdf

So then on to y ss/: which basically does a `strpos()` of each of our urls against the list we're scanning (i.e. the thing that looks like "http://url,http://otherurl". She then counts each (#:') result, takes the max again (|/). This means if ANY of those match, return `1b` otherwise return `0b`.

Every time these k folks show up on HN, it feels like Chuck Moore and Color Forth.

Reply on news.ycombinator.com