Topic: A diversion for the recreationally challenged.

[MP]

Glad to help.

[gskelly]

Is the lambda stuff compatible with other drafting programs such as Bricscad?

There should be no problem with lambda in Bricscad. Just did a quick copy/paste of Lee's swamp-ss-zoom-extents and it appears (visually) to work as expected in Bricscad v10. (I did not actually check the values of the extents returned).

[ronjonp]

Is the lambda stuff compatible with other drafting programs such as Bricscad?

There should be no problem with lambda in Bricscad. Just did a quick copy/paste of Lee's swamp-ss-zoom-extents and it appears (visually) to work as expected in Bricscad v10. (I did not actually check the values of the extents returned).

Good to know   I wonder why while isn't used more since it seems to usually be the winner in speed "challenges".

[alanjt]

Is the lambda stuff compatible with other drafting programs such as Bricscad?

There should be no problem with lambda in Bricscad. Just did a quick copy/paste of Lee's swamp-ss-zoom-extents and it appears (visually) to work as expected in Bricscad v10. (I did not actually check the values of the extents returned).

Good to know   I wonder why while isn't used more since it seems to usually be the winner in speed "challenges".

I thought repeat was faster than while.

[ronjonp]

Is the lambda stuff compatible with other drafting programs such as Bricscad?

There should be no problem with lambda in Bricscad. Just did a quick copy/paste of Lee's swamp-ss-zoom-extents and it appears (visually) to work as expected in Bricscad v10. (I did not actually check the values of the extents returned).

Good to know   I wonder why while isn't used more since it seems to usually be the winner in speed "challenges".

I thought repeat was faster than while.

Is it? I could be waaaaaay off again 

[MP]

hmmm ... logic suggests a while loop would be slower because tests are evaluated at the lisp (interpreted) level, whereas in a repeat loop they're evaluated at the executable level; most likely using a register for the counter value. Could be over thinking it tho, been a long day.

[JohnK]

repeat, foreach, while. In that order.

[gile]

Hi,

As Michael said, logicaly repeat sould be faster than while, but it appears to me that most of the time while seems to be faster.
Anyway, a while statement speed depends on what is evaluated in the conditional expression.

Here's an example with three implementations of the classical ss2lst routine which returns a list of ename from a selection set.
The first one (ss2lst_repeat) uses repeat with the length of the selection set.
The second one (ss2lst_while1) uses while the same way as repeat evaluating a counter
The third one (ss2lst_while2) uses while in an optimised way: evaluating (ssname ss n) as condition, this have to be evaluated in every case.
I don't include foreach, neither mapcar, which only operate on lists.

Code: [Select]

(defun ss2lst_repeat (ss / n lst)
  (repeat (setq n (sslength ss))
    (setq lst (cons (ssname ss (setq n (1- n))) lst))
  )
  lst
)

(defun ss2lst_while1 (ss / n lst)
  (setq n (sslength ss))
  (while (< 0 n)
    (setq lst (cons (ssname ss (setq n (1- n))) lst))
  )
  lst
)

(defun ss2lst_while2 (ss / n ent lst)
  (setq n -1)
  (while (setq ent (ssname ss (setq n (1+ n))))
    (setq lst (cons ent lst))
  )
  (reverse lst)
)

_$ (sslength (setq ss (ssget)))
10000
_$ (benchmark '((ss2lst_repeat ss) (ss2lst_while1 ss) (ss2lst_while2 ss)))
Benchmarking ..........Elapsed milliseconds / relative speed for 128 iteration(s):

    (SS2LST_WHILE2 SS).....1060 / 1.35 <fastest>
    (SS2LST_WHILE1 SS).....1248 / 1.15
    (SS2LST_REPEAT SS).....1435 / 1.00 <slowest>

[ElpanovEvgeniy]

Code: [Select]

(sslength (setq ss (ssget "_x")))
(benchmark '((ss2lst_repeat ss) (ss2lst_while1 ss)(ss2lst_while2 ss) (ACET-SS-TO-LIST ss)))


Code: [Select]

_$

10000
_$
Benchmarking ...........Elapsed milliseconds / relative speed for 256 iteration(s):

    (ACET-SS-TO-LIST SS).....1389 / 1.85 <fastest>
    (SS2LST_WHILE2 SS).......1888 / 1.36
    (SS2LST_WHILE1 SS).......2278 / 1.13
    (SS2LST_REPEAT SS).......2574 / 1 <slowest>

 
_$

[Kerry]

 

(ACET-SS-TO-LIST SS).....1389 / 1.85 <fastest>

[ElpanovEvgeniy]

 

(ACET-SS-TO-LIST SS).....1389 / 1.85 <fastest>

It is easy to explain - the program is already compiled versions of the...

[ElpanovEvgeniy]

after compilation:

Code: [Select]

(defun c:test (/ ss)
 (sslength (setq ss (ssget "_x")))
 (benchmark
  '((ss2lst_repeat ss) (ss2lst_while1 ss) (ss2lst_while2 ss) (ACET-SS-TO-LIST ss))
 )
 (princ)
)

Code: [Select]

(LOAD "C:/Users/EElpanov/Desktop/test/test.VLX") nil
Elapsed milliseconds / relative speed for 512 iteration(s):
    (SS2LST_REPEAT SS).......1155 / 2.42 <fastest>
    (SS2LST_WHILE1 SS).......1170 / 2.39
    (SS2LST_WHILE2 SS).......1389 / 2.01
    (ACET-SS-TO-LIST SS).....2792 / 1 <slowest>

[Kerry]

That sounds better

The ACET-SS-TO-LIST jumps through a lot of hoops
Checks if ss
checks (EQUAL (TYPE SS) PICKSET)
checks (> (SSLENGTH SS) 0)

Then runs a repeat on the ... (SSNAME SS n)

cons a list

Reverse the list when finished.

[Lee Mac]

Wow - I didn't realise compilation made that much of a difference....

[ronjonp]

So we can say while beats repeat except after c and when compiled?