The APL Orchard

rabbitgrowth

3:13 AM

@doug Thanks for this, but that's a different section (3.7; page 108 is in 3.6)

doug

3:33 AM

How many page 108’s are there?

doug

3:53 AM

Oh, you’re asking why the picture on 108 doesn’t line up with the picture on 117? It’s just a different AST as noted.

Ha! I only just noticed that the trees in the correction are just completely different than those in his thesis. I only pointed out that the after doesn’t match the before.

rabbitgrowth

@doug No, the apparent mismatch is between the "Before" tree on page 108 and the "before" vectors on page 109. Sorry I didn't make that clearer.

doug

4:22 AM

Between the drawing and the data? I would have hoped the correction would address something like this but it’s not something I checked. I only noticed the trees in the thesis reflect the action described. Maybe drop him a note..

rabbitgrowth

Well, between the "After" tree and "after" vectors as well, actually. I'm guessing it's just an error.

@doug Yes

doug

The after tree is just wrong. That was the point of the correction.

I obviously meant “don’t reflect”…

rabbitgrowth

The correction is to a different part (page 117, in section 3.7) than the one I'm struggling with (page 108-109, in section 3.6)

doug

4:40 AM

Yeah. It’s a little tough to follow since the correction looks like neither of these diagrams.

doug

4:56 AM

Just from the parent vector alone I get a tree of this shape:

a  e
│  ├─────┬──┐
b  f     l  o
│  ├──┐  │  │
c  g  h  m  p
│     │  │  ├──┬──┐
d     i  n  q  s  t
      │     │
      j     r
      │
      k

So I’d guess there are errors.

rabbitgrowth

5:14 AM

What did you use to make that diagram?

Yeah, that's the same as the "After" tree on page 108, except every A─V is replaced with just V (judging from the type vector)

doug

5:29 AM

I did it in K. Note that the parent vector as given is no longer in DFS preorder.

I have a translation in APL but only of the plotting itself not with the other tools.

It might be a bit cryptic but it’s pretty constructive compared to the recursive version Aaron has.

FWIW, a various code sketches here.

rabbitgrowth

@doug Nice, I'll try to read this

doug

Oops! There might be a bug in the APL version. I fixed one a while back in K but don’t know if it made it to the APL. I’ll check a bit later.

It shouldn’t be awful, but still.

doug

6:19 AM

Maybe it’s fine. I think I’m remembering a different bug.

doug

10:05 AM

In case it's useful, I mindlessly copied all the data about trees from the thesis and [rendered them here](https://tinyurl.com/3zjuhyvu). I've noted the parent vector actually used for rendering but output some other arrays like a depth vector before processing or a raw parent vector before resorting into DFS preorder.

A few look funny which may well be my bugs. I'll have a closer look at some point.

In any event, take it for what it's worth.

B. Wilson

1:07 PM

@Adám Another post to my blog, this time reacting to @PaulMansour: blog.wilsonb.com/posts/….

Transcript for