Re: Coding challenge

[Ben Rubinstein]

I'm interested that nobody went recursive.  My first solution (I read Mark's 
email and stopped reading until I'd done a version) was


command testChange
   local t
   put empty into t
   repeat 10 times
  get random(99)
  put format(to make %d:%s\n, it, makeChange(it)) after t
   end repeat
   put t
end testChange

function makeChange iAmount, tChangeSoFar
   constant kCoins = 50 20 10 5 2 1
   if iAmount  1 then return tChangeSoFar
   repeat for each word c in kCoins
  if iAmount = c then
 return makeChange(iAmount - c, tChangeSoFar  c)
  end if
   end repeat
   return error
end makeChange

(note the especially crafty thing here: that by citing my test function, I've 
justified leaving a leading space in my output...).


So far, effectively like the rest - I don't think the recursion is better, I 
was just surprised that nobody else naturally went that way.


When I replaced the coins list with the unusual denominations of Craig$, 6 
came out as 4+1+1, as you'd expect.  Here I think is where recursion makes 
things better, because it's easy to adjust:


function makeChange iAmount, tChangeSoFar
   local tCandidates
   constant kCoins = 40 30 10 4 3 1 -- 50 20 10 5 2 1
   if iAmount  1 then return tChangeSoFar
   put empty into tCandidates
   repeat for each word c in kCoins
  if iAmount  c then next repeat
  put makeChange(iAmount - c, c)  return after tCandidates
   end repeat
   sort lines of tCandidates ascending numeric by the number of words in each
   return tChangeSoFar  first line of tCandidates
end makeChange

With these small changes this works, and I'm confident returns the optimal 
solution - but it gets shockingly slow for values above about 20.


The first reason it was so slow is because I'm being lazy above - allowing the 
code to try options in the wrong order, as each call tries the full set of 
coins again.  I tried speeding it up by making it having found a coin it can 
use, try with that and the next one down, but not all possibles.  I'm 
reasonably confident that this produced the right result for all cases with 
Craig's currency - but I wasn't absolutely sure, and it's certainly not a 
general solution, eg asked to make 27 out of 40 30 23 20 9 1 - admittedly a 
really crazy currency - it would offer 23 1 1 1 1 rather than the optimal 9 
9 9.


So in the end I had to rewrite it more substantially:

command testChange
   constant kCoins = 40 30 10 4 3 1 -- 50 20 10 5 2 1
   local t
   put empty into t
   repeat 10 times
  get random(99)
  put format(to make %d:%s\n, it, makeChange(it, kCoins)) after t
   end repeat
   put t
end testChange


function makeChange iAmount, tCoins, tChangeSoFar
   local tCandidates, i, c, j
   if iAmount  1 then return tChangeSoFar

   -- trim coins that couldn't possibly be used
   repeat with i = number of words in tCoins down to 1
  if word i of tCoins  iAmount then delete word i of tCoins
   end repeat

   -- find possible combinations
   put empty into tCandidates
   repeat with i = 1 to number of words in tCoins
  put word i of tCoins into c
  put makeChange(iAmount - c, word i to -1 of tCoins, c) \
 return after tCandidates
   end repeat

   -- return shortest combination
   sort lines of tCandidates ascending numeric by the number of words in each
   return tChangeSoFar  first line of tCandidates
end makeChange

This definitely returns the optimal combination.  It's quite slow (fractions 
of a second, but noticeable fractions).  But I feel entirely justified in 
playing LiveCode's joker:


local saAmountCoins2Change

function makeChange iAmount, tCoins, tChangeSoFar
   local tCandidates, i, c, j
   if iAmount  1 then return tChangeSoFar

   -- have we done this before?
   get saAmountCoins2Change[iAmount, tCoins]
   if it  empty then return tChangeSoFar  it

   -- trim coins that couldn't possibly be used
   repeat with i = number of words in tCoins down to 1
  if word i of tCoins  iAmount then delete word i of tCoins
   end repeat

   -- find possible combinations
   put empty into tCandidates
   repeat with i = 1 to number of words in tCoins
  put word i of tCoins into c
  put makeChange(iAmount - c, word i to -1 of tCoins, c) \
 return after tCandidates
   end repeat

   -- return shortest combination
   sort lines of tCandidates ascending numeric by the number of words in each
   put first line of tCandidates into saAmountCoins2Change[iAmount, tCoins]
   return tChangeSoFar  first line of tCandidates
end makeChange

Interestingly, even if the array is emptied before each run of tests, this 
changes the typical time for a run of tests from approx 1 second for the ten, 
to approx 7 milliseconds for the ten.  Of course, if the array isn't emptied, 
after a few runs it approaches zero.  (If the array is emptied before each 
single test, which could be described as fairer, it averages around 20 
milliseconds for ten tests).


But of course, all of that is only necessary if we move to 

Re: Coding challenge

[dunbarx]

Paul.


As six pennies. As long as you have a 1, you should be OK.




Craig Newman




-Original Message-
From: Paul D. DeRocco pdero...@ix.netcom.com
To: 'How to use LiveCode' use-livecode@lists.runrev.com
Sent: Thu, Jan 31, 2013 12:57 am
Subject: RE: Coding challenge


 From: Mark Wieder
 
  Now how would you do it if the available coin values were:
 
  40,30,10,4,3,1
 
  That's a more interesting problem, but probably a less 
  interesting coding
  test, because I think it would involve a more brute force 
  approach, less
  elegance.
 
 I'm missing something. Why would that be different?

How would you represent 6?

-- 

Ciao,   Paul D. DeRocco
Paulmailto:pdero...@ix.netcom.com 


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

 

___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[Alex Tweedly]

The question was

Determine the minimum number of coins for change.
so the correct answer here would be 2 coins (3+3) rather than 6 coins 
(1+1+...)


That's what makes this a more challenging case, but probably without as 
elegant an answer ...


-- Alex.

On 31/01/2013 13:54, dunb...@aol.com wrote:

Paul.


As six pennies. As long as you have a 1, you should be OK.




Craig Newman




-Original Message-
From: Paul D. DeRocco pdero...@ix.netcom.com
To: 'How to use LiveCode' use-livecode@lists.runrev.com
Sent: Thu, Jan 31, 2013 12:57 am
Subject: RE: Coding challenge



From: Mark Wieder


Now how would you do it if the available coin values were:
 40,30,10,4,3,1
That's a more interesting problem, but probably a less
interesting coding
test, because I think it would involve a more brute force
approach, less
elegance.

I'm missing something. Why would that be different?

How would you represent 6?




___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[Colin Holgate]

Here's my one:

on makeChange Amt
  answer 100-Amt
end makeChange

Oh wait, that's for showing the maximum number of coins…



___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[Mark Wieder]

Alex-

Thursday, January 31, 2013, 6:53:36 AM, you wrote:

 The question was
 Determine the minimum number of coins for change.
 so the correct answer here would be 2 coins (3+3) rather than 6 coins
 (1+1+...)

 That's what makes this a more challenging case, but probably without as
 elegant an answer ...

In any of the submitted examples, replacing the string
50,20,10,5,2,1 with 40,30,10,4,3,1 does the right string. No more
challenging than the original question.

And yes, the minimum number of coins to represent 6 in this case is
3+3 and the next in line would be 4+1+1. And then six coins.

-- 
-Mark Wieder
 mwie...@ahsoftware.net


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[dunbarx]

Hi.


I sped-read your comment, and answered generically. You might say that I 
answered for the number 4, with only 20,10,5,1 as options.



But all the several solutions sent to Mark ran from high denominations to low. 
So two 3's would appear before it was ever an issue to load a bunch of 1's. 
The minimum number of coins was always the result.



Craig Newman




-Original Message-
From: Alex Tweedly a...@tweedly.net
To: use-livecode use-livecode@lists.runrev.com
Sent: Thu, Jan 31, 2013 9:54 am
Subject: Re: Coding challenge


The question was
 Determine the minimum number of coins for change.
so the correct answer here would be 2 coins (3+3) rather than 6 coins 
(1+1+...)

That's what makes this a more challenging case, but probably without as 
elegant an answer ...

-- Alex.

On 31/01/2013 13:54, dunb...@aol.com wrote:
 Paul.


 As six pennies. As long as you have a 1, you should be OK.




 Craig Newman




 -Original Message-
 From: Paul D. DeRocco pdero...@ix.netcom.com
 To: 'How to use LiveCode' use-livecode@lists.runrev.com
 Sent: Thu, Jan 31, 2013 12:57 am
 Subject: RE: Coding challenge


 From: Mark Wieder

 Now how would you do it if the available coin values were:
  40,30,10,4,3,1
 That's a more interesting problem, but probably a less
 interesting coding
 test, because I think it would involve a more brute force
 approach, less
 elegance.
 I'm missing something. Why would that be different?
 How would you represent 6?



___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

 
___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[dunbarx]

Colin.


At least you are true-to-form, in providing a minimum line solution. Feels like 
old times. How about:



on x

end x




Doesn't do much that is useful, true, but is the shortest possible handler.


Craig


-Original Message-
From: Colin Holgate co...@verizon.net
To: How to use LiveCode use-livecode@lists.runrev.com
Sent: Thu, Jan 31, 2013 10:05 am
Subject: Re: Coding challenge


Here's my one:

on makeChange Amt
  answer 100-Amt
end makeChange

Oh wait, that's for showing the maximum number of coins…



___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

 
___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[Colin Holgate]

This is one case where Lingo (the main language used in Director) would be 
better, you don't have to include the end line. So this would be just as good 
(or bad, depending on how you look at it):

on x


On Jan 31, 2013, at 12:44 PM, dunb...@aol.com wrote:

 
 on x
 
 end x
 
 
 
 
 Doesn't do much that is useful, true, but is the shortest possible handler.

___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

RE: Coding challenge

[Paul D. DeRocco]

 Mark Wieder
 
 In any of the submitted examples, replacing the string
 50,20,10,5,2,1 with 40,30,10,4,3,1 does the right string. No more
 challenging than the original question.
 
 And yes, the minimum number of coins to represent 6 in this case is
 3+3 and the next in line would be 4+1+1. And then six coins.

Yes, and the examples so far would render 6 as 4+1+1, not 3+3, which is the
wrong answer.

-- 

Ciao,   Paul D. DeRocco
Paulmailto:pdero...@ix.netcom.com 


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[Robert Sneidar]

Someone should post Jacques as an example of how coding in Livecode is so much 
more compact. 

Bob


On Jan 30, 2013, at 8:45 PM, J. Landman Gay wrote:

 On 1/30/13 10:29 PM, Mark Wieder wrote:
 
 And everyone but me seemed to think of listing the number of coin
 types. I got lazy and just repeated the coin if necessary.
 
 
 Regardless, all the solutions are  way more readable than the web page 
 examples.
 
 -- 
 Jacqueline Landman Gay | jac...@hyperactivesw.com
 HyperActive Software   | http://www.hyperactivesw.com
 
 ___
 use-livecode mailing list
 use-livecode@lists.runrev.com
 Please visit this url to subscribe, unsubscribe and manage your subscription 
 preferences:
 http://lists.runrev.com/mailman/listinfo/use-livecode


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[Robert Sneidar]

Ah, but you DID remember it! Which is more that I seem capable of these days. 
:-)

Bob


On Jan 30, 2013, at 8:56 PM, J. Landman Gay wrote:

 On 1/30/13 10:38 PM, dunb...@aol.com wrote:
 I think, as usual, Jacque's is the most original.
 
 Thanks, but it's really just a variation on calculating digital time (hours, 
 mins, secs) from seconds. I didn't think up that one.
 
 -- 
 Jacqueline Landman Gay | jac...@hyperactivesw.com
 HyperActive Software   | http://www.hyperactivesw.com
 
 ___
 use-livecode mailing list
 use-livecode@lists.runrev.com
 Please visit this url to subscribe, unsubscribe and manage your subscription 
 preferences:
 http://lists.runrev.com/mailman/listinfo/use-livecode


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[dunbarx]

Ah.


Paul is right, the descending scheme finds 4,1,1 instead of 3,3, because 
since the 4 is used first this leaves only 1's available for the rest of the 
work. 



Ah.



This is now a more interesting challenge. It requires that the nextmost lower 
value be considered as a candidate. And maybe even lower nextmost ones?




Craig.


-Original Message-
From: Paul D. DeRocco pdero...@ix.netcom.com
To: 'How to use LiveCode' use-livecode@lists.runrev.com
Sent: Thu, Jan 31, 2013 12:55 pm
Subject: RE: Coding challenge


 Mark Wieder
 
 In any of the submitted examples, replacing the string
 50,20,10,5,2,1 with 40,30,10,4,3,1 does the right string. No more
 challenging than the original question.
 
 And yes, the minimum number of coins to represent 6 in this case is
 3+3 and the next in line would be 4+1+1. And then six coins.

Yes, and the examples so far would render 6 as 4+1+1, not 3+3, which is the
wrong answer.

-- 

Ciao,   Paul D. DeRocco
Paulmailto:pdero...@ix.netcom.com 


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

 
___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[Robert Sneidar]

4 and 2 1's? Isn't the test to produce the  result with the least amount of 
coins? This would still work. The idea is to try and fit the largest coin value 
into the remainder each time, until the remainder is zero. 

Bob


On Jan 31, 2013, at 5:54 AM, dunb...@aol.com wrote:

 Paul.
 
 
 As six pennies. As long as you have a 1, you should be OK.
 
 
 
 
 Craig Newman
 
 
 
 
 -Original Message-
 From: Paul D. DeRocco pdero...@ix.netcom.com
 To: 'How to use LiveCode' use-livecode@lists.runrev.com
 Sent: Thu, Jan 31, 2013 12:57 am
 Subject: RE: Coding challenge
 
 
 From: Mark Wieder
 
 Now how would you do it if the available coin values were:
 
40,30,10,4,3,1
 
 That's a more interesting problem, but probably a less 
 interesting coding
 test, because I think it would involve a more brute force 
 approach, less
 elegance.
 
 I'm missing something. Why would that be different?
 
 How would you represent 6?
 
 -- 
 
 Ciao,   Paul D. DeRocco
 Paulmailto:pdero...@ix.netcom.com 
 
 
 ___
 use-livecode mailing list
 use-livecode@lists.runrev.com
 Please visit this url to subscribe, unsubscribe and manage your subscription 
 preferences:
 http://lists.runrev.com/mailman/listinfo/use-livecode
 
 
 
 ___
 use-livecode mailing list
 use-livecode@lists.runrev.com
 Please visit this url to subscribe, unsubscribe and manage your subscription 
 preferences:
 http://lists.runrev.com/mailman/listinfo/use-livecode


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[Robert Sneidar]

OIC now. 3  3 is the best answer but  the algorithm produces 4 + 1 + 1. Well I 
think the issue here is that currency is never (or almost never) designed this 
way. No one would make a 3 dollar bill coincidentally with a 4 dollar bill. 
There would be no practical reason to. 

And yet I remember for a time the USA produced a one dollar bill and a two 
dollar bill. So never say never! :-)

It seems that when creating currency values, one of the overriding principles 
OUGHT to be that each smaller value ought to divide evenly into all the larger 
values. I'm sure this is what was intended with American currency when it was 
first created in it's present form. Otherwise making change becomes rather 
tedious. 

Bob


On Jan 30, 2013, at 9:56 PM, Paul D. DeRocco wrote:

 From: Mark Wieder
 
 Now how would you do it if the available coin values were:
 
40,30,10,4,3,1
 
 That's a more interesting problem, but probably a less 
 interesting coding
 test, because I think it would involve a more brute force 
 approach, less
 elegance.
 
 I'm missing something. Why would that be different?
 
 How would you represent 6?
 
 -- 
 
 Ciao,   Paul D. DeRocco
 Paulmailto:pdero...@ix.netcom.com 
 
 
 ___
 use-livecode mailing list
 use-livecode@lists.runrev.com
 Please visit this url to subscribe, unsubscribe and manage your subscription 
 preferences:
 http://lists.runrev.com/mailman/listinfo/use-livecode


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[J. Landman Gay]

On 1/31/13 9:43 AM, Mark Wieder wrote:

Alex-

Thursday, January 31, 2013, 6:53:36 AM, you wrote:


The question was

Determine the minimum number of coins for change.

so the correct answer here would be 2 coins (3+3) rather than 6 coins
(1+1+...)



That's what makes this a more challenging case, but probably without as
elegant an answer ...


In any of the submitted examples, replacing the string
50,20,10,5,2,1 with 40,30,10,4,3,1 does the right string. No more
challenging than the original question.

And yes, the minimum number of coins to represent 6 in this case is
3+3 and the next in line would be 4+1+1. And then six coins.



Actually, after replacing the coin values in my original solution, 
change for 6 comes out:


4=1
1=2

Which is the correct change but not the minimum number of coins.

--
Jacqueline Landman Gay | jac...@hyperactivesw.com
HyperActive Software   | http://www.hyperactivesw.com

___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[Mark Wieder]

Paul D. DeRocco pderocco@... writes:

 Yes, and the examples so far would render 6 as 4+1+1, not 3+3, which is the
 wrong answer.
 

Ah. Right you are.

-- 
 Mark Wieder
 mwie...@ahsoftware.net


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[Dave Cragg]

On 31 Jan 2013, at 05:40, Paul D. DeRocco pdero...@ix.netcom.com wrote:

 Now how would you do it if the available coin values were:
 
   40,30,10,4,3,1
 
 That's a more interesting problem, but probably a less interesting coding
 test, because I think it would involve a more brute force approach, less
 elegance.

Right. Not only finding the least number of coins, but also checking that the 
available coins can be used to create the desired total. The presence of a 
1-unit coin should make that so, so long as you check the desired total is an 
integer. Not all solutions checked for that. Tut tut! :-)

Dave
___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

RE: Coding challenge

[Paul D. DeRocco]

 From: Robert Sneidar
 
 OIC now. 3  3 is the best answer but  the algorithm produces 
 4 + 1 + 1. Well I think the issue here is that currency is 
 never (or almost never) designed this way. No one would make 
 a 3 dollar bill coincidentally with a 4 dollar bill. There 
 would be no practical reason to. 
 
 And yet I remember for a time the USA produced a one dollar 
 bill and a two dollar bill. So never say never! :-)
 
 It seems that when creating currency values, one of the 
 overriding principles OUGHT to be that each smaller value 
 ought to divide evenly into all the larger values. I'm sure 
 this is what was intended with American currency when it was 
 first created in it's present form. Otherwise making change 
 becomes rather tedious. 

Actually, there was a brief time (1875-1888) when we had both a quarter and
a 20-cent coin. So the algorithm would make 40 cents out of 25+10+5, instead
of 20+20.

-- 

Ciao,   Paul D. DeRocco
Paulmailto:pdero...@ix.netcom.com 


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[dunbarx]

We also had both a nickel and a half dime together for several years. I cannot 
figure out how to handle that.


Craig Newman



-Original Message-
From: Paul D. DeRocco pdero...@ix.netcom.com
To: 'How to use LiveCode' use-livecode@lists.runrev.com
Sent: Thu, Jan 31, 2013 3:02 pm
Subject: RE: Coding challenge


 From: Robert Sneidar
 
 OIC now. 3  3 is the best answer but  the algorithm produces 
 4 + 1 + 1. Well I think the issue here is that currency is 
 never (or almost never) designed this way. No one would make 
 a 3 dollar bill coincidentally with a 4 dollar bill. There 
 would be no practical reason to. 
 
 And yet I remember for a time the USA produced a one dollar 
 bill and a two dollar bill. So never say never! :-)
 
 It seems that when creating currency values, one of the 
 overriding principles OUGHT to be that each smaller value 
 ought to divide evenly into all the larger values. I'm sure 
 this is what was intended with American currency when it was 
 first created in it's present form. Otherwise making change 
 becomes rather tedious. 

Actually, there was a brief time (1875-1888) when we had both a quarter and
a 20-cent coin. So the algorithm would make 40 cents out of 25+10+5, instead
of 20+20.

-- 

Ciao,   Paul D. DeRocco
Paulmailto:pdero...@ix.netcom.com 


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

 
___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[Mark Wieder]

 dunbarx@... writes:

 
 We also had both a nickel and a half dime together for several years. I cannot
figure out how to handle that.

Don't sweat it.
You put them both in the same coin slot.
The only time you need to worry about it is when the slot runs dry.
For the same reason, you don't need to check the date on a nickel before handing
it out in change.

-- 
 Mark Wieder
 mwie...@ahsoftware.net





___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[dunbarx]

Mark.


For the same reason, you don't need to check the date on a nickel before 
handing
it out in change.
Ha.


Please hand me a nice, unworn 1926S nickel in my change. Or maybe a 1913 
liberty?



Craig




-Original Message-
From: Mark Wieder mwie...@ahsoftware.net
To: use-livecode use-livecode@lists.runrev.com
Sent: Thu, Jan 31, 2013 4:29 pm
Subject: Re: Coding challenge


 dunbarx@... writes:

 
 We also had both a nickel and a half dime together for several years. I cannot
figure out how to handle that.

Don't sweat it.
You put them both in the same coin slot.
The only time you need to worry about it is when the slot runs dry.
For the same reason, you don't need to check the date on a nickel before handing
it out in change.

-- 
 Mark Wieder
 mwie...@ahsoftware.net





___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

 

___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[Monte Goulding]

It would depend on what's in your wallet ;-)

On 31/01/2013, at 2:16 PM, Mark Wieder mwie...@ahsoftware.net wrote:

 ...and the xtalk implementation would be...

--
Monte Goulding

M E R Goulding - software development services
mergExt - There's an external for that!





___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[Terry Judd]

OK - I'm usually crap at these things but I'll have a go...

function getChange pTotal
   put 50,20,10,5,2,1 into tCoins
   local tCounts
   put 0 into tValue
   repeat for each item tCoin in tCoins
  put trunc(pTotal/tCoin) into tValue
  put tValue  space after tCounts
  subtract tCoin*tValue from pTotal
   end repeat
   return tCounts
end getChange

Terry...

On 31/01/2013, at 02:16 PM, Mark Wieder wrote:

 Today's algorithm challenge:
 
 Determine the minimum number of coins for change.
 
 Given any number between 1 and 99, determine how to give change with
 the minimum number of coins. You can assume that the coins are 1c, 2c,
 5c, 10c, 20c and 50c.
 
 Various solutions here in Scala, java, C#, Groovy...
 http://java.dzone.com/articles/thursday-code-puzzler-change
 
 ...and the xtalk implementation would be...
 
 -- 
 -Mark Wieder
 mwie...@ahsoftware.net
 
 
 ___
 use-livecode mailing list
 use-livecode@lists.runrev.com
 Please visit this url to subscribe, unsubscribe and manage your subscription 
 preferences:
 http://lists.runrev.com/mailman/listinfo/use-livecode
 

Dr Terry Judd
Senior Lecturer in Medical Education
Medical Eduction Unit
Faculty of Medicine, Dentistry  Health Sciences
The University of Melbourne





___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[J. Landman Gay]

On 1/30/13 9:16 PM, Mark Wieder wrote:

Today's algorithm challenge:

Determine the minimum number of coins for change.

Given any number between 1 and 99, determine how to give change with
the minimum number of coins. You can assume that the coins are 1c, 2c,
5c, 10c, 20c and 50c.

Various solutions here in Scala, java, C#, Groovy...
http://java.dzone.com/articles/thursday-code-puzzler-change

...and the xtalk implementation would be...




Didn't test all possible values, but so far this works:

on makeChange pAmt
  repeat for each item i in 50,20,10,5,2,1
put i = pAmt div i  cr after tList
put pAmt mod i into pAmt
  end repeat
  filter tList without *=0
  put tList into fld change
end makeChange

--
Jacqueline Landman Gay | jac...@hyperactivesw.com
HyperActive Software   | http://www.hyperactivesw.com

___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[Mark Wieder]

Terry-

Wednesday, January 30, 2013, 7:56:49 PM, you wrote:

 OK - I'm usually crap at these things but I'll have a go...

 function getChange pTotal
put 50,20,10,5,2,1 into tCoins
local tCounts
put 0 into tValue
repeat for each item tCoin in tCoins
   put trunc(pTotal/tCoin) into tValue
   put tValue  space after tCounts
   subtract tCoin*tValue from pTotal
end repeat
return tCounts
 end getChange

That'll do it.
Running your implementation I get a number for each coin type.
As in 74 = 1 1 0 0 2 0

-- 
-Mark Wieder
 mwie...@ahsoftware.net


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[Mark Wieder]

Jacque-

Wednesday, January 30, 2013, 7:59:42 PM, you wrote:

 on makeChange pAmt
repeat for each item i in 50,20,10,5,2,1
  put i = pAmt div i  cr after tList
  put pAmt mod i into pAmt
end repeat
filter tList without *=0
put tList into fld change
 end makeChange

Mod - nicely done.
I hadn't thought of the solution of sorting by coin type.
So I get 74 =
50=1
20=1
2=2

-- 
-Mark Wieder
 mwie...@ahsoftware.net


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[Mark Wieder]

Here's the one I came up with:

function makeChange pValue
local tChange

set the itemdelimiter to comma
repeat for each item tCoin in 50,20,10,5,2,1
repeat while pValue = tCoin
put tCoin  comma after tChange
subtract tCoin from pValue
end repeat
end repeat
return char 1 to -2 of tChange
end makeChange

And for 74 I get 50,20,2,2

-- 
-Mark Wieder
 mwie...@ahsoftware.net


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[Paul Hibbert]

And here's my simple logical non-mathematician version!

function coinChange pChange
   repeat for each item x in 50, 20, 10, 5, 2, 1
  repeat while pChange = x
 subtract x from pChange
 add 1 to y
  end repeat
  if y  0 then put y  x  x  c  cr after tResult
  put 0 into y
   end repeat
   return tResult
end coinChange

Paul

On 2013-01-30, at 7:16 PM, Mark Wieder wrote:

 Today's algorithm challenge:
 
 Determine the minimum number of coins for change.
 
 Given any number between 1 and 99, determine how to give change with
 the minimum number of coins. You can assume that the coins are 1c, 2c,
 5c, 10c, 20c and 50c.
 
 Various solutions here in Scala, java, C#, Groovy...
 http://java.dzone.com/articles/thursday-code-puzzler-change
 
 ...and the xtalk implementation would be...
 
 -- 
 -Mark Wieder
 mwie...@ahsoftware.net
 
 
 ___
 use-livecode mailing list
 use-livecode@lists.runrev.com
 Please visit this url to subscribe, unsubscribe and manage your subscription 
 preferences:
 http://lists.runrev.com/mailman/listinfo/use-livecode


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[Mark Wieder]

Paul-

Wednesday, January 30, 2013, 8:21:36 PM, you wrote:

 And here's my simple logical non-mathematician version!

 function coinChange pChange
repeat for each item x in 50, 20, 10, 5, 2, 1
   repeat while pChange = x
  subtract x from pChange
  add 1 to y
   end repeat
   if y  0 then put y  x  x  c  cr after tResult
   put 0 into y
end repeat
return tResult
 end coinChange

You did the same nested repeat thing I did.
And everyone but me seemed to think of listing the number of coin
types. I got lazy and just repeated the coin if necessary.

-- 
-Mark Wieder
 mwie...@ahsoftware.net


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[dunbarx]

Mark:


on mouseup
  ask Enter Value
put it into temp
put 50,20,10,5,2,1 into coins
put 1 into tCount
repeat until temp = 0
  if item tCount of coins = temp then
put item tCount of coins  , after tChange
subtract item tCount of coins from temp
else
add 1 to tCount
end if
end repeat
answer tChange
end mouseup


Craig
___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[dunbarx]

I think, as usual, Jacque's is the most original.


Craig


on makeChange pAmt
   repeat for each item i in 50,20,10,5,2,1
 put i = pAmt div i  cr after tList
 put pAmt mod i into pAmt
   end repeat
   filter tList without *=0
   put tList into fld change
end makeChange





-Original Message-
From: J. Landman Gay jac...@hyperactivesw.com
To: How to use LiveCode use-livecode@lists.runrev.com
Sent: Wed, Jan 30, 2013 11:00 pm
Subject: Re: Coding challenge


On 1/30/13 9:16 PM, Mark Wieder wrote:
 Today's algorithm challenge:

 Determine the minimum number of coins for change.

 Given any number between 1 and 99, determine how to give change with
 the minimum number of coins. You can assume that the coins are 1c, 2c,
 5c, 10c, 20c and 50c.

 Various solutions here in Scala, java, C#, Groovy...
 http://java.dzone.com/articles/thursday-code-puzzler-change

 ...and the xtalk implementation would be...



Didn't test all possible values, but so far this works:

on makeChange pAmt
   repeat for each item i in 50,20,10,5,2,1
 put i = pAmt div i  cr after tList
 put pAmt mod i into pAmt
   end repeat
   filter tList without *=0
   put tList into fld change
end makeChange

-- 
Jacqueline Landman Gay | jac...@hyperactivesw.com
HyperActive Software   | http://www.hyperactivesw.com

___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

 
___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[J. Landman Gay]

On 1/30/13 10:29 PM, Mark Wieder wrote:


And everyone but me seemed to think of listing the number of coin
types. I got lazy and just repeated the coin if necessary.



Regardless, all the solutions are  way more readable than the web page 
examples.


--
Jacqueline Landman Gay | jac...@hyperactivesw.com
HyperActive Software   | http://www.hyperactivesw.com

___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[Mark Wieder]

Craig-

Wednesday, January 30, 2013, 8:31:54 PM, you wrote:

 on mouseup
   ask Enter Value
 put it into temp
 put 50,20,10,5,2,1 into coins
 put 1 into tCount
 repeat until temp = 0
   if item tCount of coins = temp then
 put item tCount of coins  , after tChange
 subtract item tCount of coins from temp
 else
 add 1 to tCount
 end if
 end repeat
 answer tChange
 end mouseup

OK. You went with my approach as well - listing the coins as values
rather than by number. I'll have to subtract a point for not cleaning
up the trailing comma, though.

-- 
-Mark Wieder
 mwie...@ahsoftware.net


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[Monte Goulding]

On 31/01/2013, at 2:22 PM, Monte Goulding mo...@sweattechnologies.com wrote:

 It would depend on what's in your wallet ;-)
 
 On 31/01/2013, at 2:16 PM, Mark Wieder mwie...@ahsoftware.net wrote:
 
 ...and the xtalk implementation would be...


on mouseUp
   put random(99) into tChange
   put tChangecrmakeChange(tChange)
end mouseUp

function makeChange pChange
   repeat for each item tCoin in 50,20,10,5,2,1
  put pChange div tCoin into tCoins
  if tCoins  0 then
 put tCoinxtCoinscr after tReturn
  end if
  put pChange mod tCoin into pChange
   end repeat
   return tReturn
end makeChange

--
Monte Goulding

M E R Goulding - software development services
mergExt - There's an external for that!





___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[J. Landman Gay]

On 1/30/13 10:38 PM, dunb...@aol.com wrote:

I think, as usual, Jacque's is the most original.


Thanks, but it's really just a variation on calculating digital time 
(hours, mins, secs) from seconds. I didn't think up that one.


--
Jacqueline Landman Gay | jac...@hyperactivesw.com
HyperActive Software   | http://www.hyperactivesw.com

___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[dunbarx]

Mark.


That comma is there for future expansion.


Craig



-Original Message-
From: Mark Wieder mwie...@ahsoftware.net
To: How to use LiveCode use-livecode@lists.runrev.com
Sent: Wed, Jan 30, 2013 11:46 pm
Subject: Re: Coding challenge


Craig-

Wednesday, January 30, 2013, 8:31:54 PM, you wrote:

 on mouseup
   ask Enter Value
 put it into temp
 put 50,20,10,5,2,1 into coins
 put 1 into tCount
 repeat until temp = 0
   if item tCount of coins = temp then
 put item tCount of coins  , after tChange
 subtract item tCount of coins from temp
 else
 add 1 to tCount
 end if
 end repeat
 answer tChange
 end mouseup

OK. You went with my approach as well - listing the coins as values
rather than by number. I'll have to subtract a point for not cleaning
up the trailing comma, though.

-- 
-Mark Wieder
 mwie...@ahsoftware.net


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

 
___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

RE: Coding challenge

[Paul D. DeRocco]

Now how would you do it if the available coin values were:

40,30,10,4,3,1

That's a more interesting problem, but probably a less interesting coding
test, because I think it would involve a more brute force approach, less
elegance.

-- 

Ciao,   Paul D. DeRocco
Paulmailto:pdero...@ix.netcom.com 


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding challenge

[Mark Wieder]

Craig-

Wednesday, January 30, 2013, 9:08:00 PM, you wrote:

 That comma is there for future expansion.

LOL.

-- 
-Mark Wieder
 mwie...@ahsoftware.net


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

RE: Coding challenge

[Paul D. DeRocco]

 From: Mark Wieder
 
  Now how would you do it if the available coin values were:
 
  40,30,10,4,3,1
 
  That's a more interesting problem, but probably a less 
  interesting coding
  test, because I think it would involve a more brute force 
  approach, less
  elegance.
 
 I'm missing something. Why would that be different?

How would you represent 6?

-- 

Ciao,   Paul D. DeRocco
Paulmailto:pdero...@ix.netcom.com 


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding Challenge

[Nonsanity]

I'll drop my code in LC and actually RUN it now. If you want to collaborate
closer on it, I'd be happy to help. My AIM/iChat is FluffAndSuch. (Other
LCers are welcome to use that too.)

 ~ Chris Innanen
 ~ Nonsnaity


On Sat, Mar 12, 2011 at 3:39 AM, Malte Brill revolut...@derbrill.de wrote:

 Chris,

 thanks for looking into this again.

 After looking closer at Steffens version there indeed are some problems
 left. Using yours (and adapting it a bit) always returned true (but I think
 this is due to one or two typos I am looking at tonight)

 Alex: I think I was a bit overenthusiastic when I said I´d works. It works
 in 95% of all cases. Not all yet. :-D So, I will also test yours again later
 today. I need this to work on Monday morning, so long weekend will be
 long. g

 Thanks again,

 Malte
 ___
 use-livecode mailing list
 use-livecode@lists.runrev.com
 Please visit this url to subscribe, unsubscribe and manage your
 subscription preferences:
 http://lists.runrev.com/mailman/listinfo/use-livecode

___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding Challenge

[Malte Brill]

Hi Alex,

thanks for taking the time to look into this. (and not late to the party at all)

If I provide this data to your Script:

Steffen  Malte
Sascha  Malte
Fritz = Sascha

I get back

Fritz-
Malte

in the field. Which is true, but it leaves steffen untouched in this case. I 
guess what I would be looking for is something like

function testValidRelationships pPerson1,pPerson2


which returns a triplet like

false,true,false

-- can not be older,can be younger, can not be the same age.

Do you think your script can be expanded to check this?

Thanks a heap again for taking the time to look into this.

Cheers,

Malte
___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding Challenge

[Alex Tweedly]

Sorry Malte - I didn't handle  relationships, only = and .
That part of the script should be changed to be

   -- now normalize the direct relationships
   put empty into gRelate
   repeat for each line  L in tData
  if word 2 of L =  then
 put sample(word 1 of L) into t1
 put sample(word 3 of L) into t2
 put t1  t2  CR after tUnequal
  else if word 2 of L =  then
 put sample(word 1 of L) into t1
 put sample(word 3 of L) into t2
 put t2  t1  CR after tUnequal
  end if
   end repeat


You asked

function testValidRelationships pPerson1,pPerson2


which returns a triplet like

false,true,false

-- can not be older,can be younger, can not be the same age.

Do you think your script can be expanded to check this?
Yes, though I would prefer a different function / result.  You have can 
not be older - but that means either is younger or  is same and you 
know which it is.


How about
function getRelationship p1, p2
-- returns one of
-- older (p1 is older than p2)
-- younger (p1 is younger)
-- same
-- unknown  (not enough info to tell anything about their relationship)


Here it is ...

function getRelationship p1, p2
   -- returns one of
   -- older (p1 is older than p2)
   -- younger (p1 is younger)
   -- same
   put sample(p1) into p1
   put sample(p2) into p2
   if p1 = p2 then return same
   if p2 is among the lines of gOlder[p1] then return younger
   if p1 is among the lines of gOlder[p2] then return older
   return unknown
end getRelationship


-- Alex.





On 11/03/2011 09:31, Malte Brill wrote:

Hi Alex,

thanks for taking the time to look into this. (and not late to the party at all)

If I provide this data to your Script:

Steffen  Malte
Sascha  Malte
Fritz = Sascha

I get back

Fritz-
Malte

in the field. Which is true, but it leaves steffen untouched in this case. I 
guess what I would be looking for is something like

function testValidRelationships pPerson1,pPerson2


which returns a triplet like

false,true,false

-- can not be older,can be younger, can not be the same age.

Do you think your script can be expanded to check this?

Thanks a heap again for taking the time to look into this.

Cheers,

Malte
___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode




___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding Challenge

[Malte Brill]

Hi Alex,

well what I need to do is to check if the relationship I am wanting to set is 
valid or not, before I write it to the database. I might not be clear enough. 
But my goal is not really to check for existing relations, but the validity of 
the data I am about to enter into the Database. So if there is no relation at 
all, any relation is valid. Direct relations are clear and indirect relations 
need to be checked for.

Maybe I just dot understand how I would use your script in that case.

Sorry for the confusion.

Malte


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding Challenge

[Nonsanity]

Try this. I haven't tested, but the logic looks sound enough...

private function resolveRelation pPers1,pPers2,pRelation
   put sRelations[pPers1][pPers2] into relationship
   -- if the two are equal but we're testing for  or  then return false
   if relationship is = and pRelation is not = then
  return false
   -- else if the relationship is a direct match, return true
   else relationship = pRelation then
  return true
   end if
   -- otherwise, interate to see if the relationship is true
   -- note that at this point, we know the two aren't directly equal
   -- this is important because the first interation HAS to be  or  and
not = or =
   -- further interations can be = and =, but the first cannot
   get traverseList(pPers1,pPers2,pRelation)
   -- if the relationship is valid, then put it into the DB to save time
later
   if it is true then
  set sRelations[pPers1][pPers2] to pRelation
   end if
   return it
end resolveRelation

private function traverseList pPers1,pPers2,pRelation
   -- are the two people of the correct relationship and/or equal?
   if sRelations[pPers2][pPers2] = pRelation or sRelations[pPers2][pPers2] =
pRelation then
  return true
   end if
   local tOther
   put the keys of sRelations[pPers1] into tOther
   repeat for each line personB in tOther
  -- only follow branches leading to the same relationship we are
testing ( or )
  -- OR =, so long as the first iteration was  or 
  -- So if: AB and BC and C=D and DE then AE must be true
  if sRelations[pPers2][personB] = pRelation or
sRelations[pPers2][personB] = = then
  return traverseList(personB,pPers2,pRelation)
  end if
   end repeat
   return false
end traverseList


 ~ Chris Innanen
 ~ Nonsanity


On Fri, Mar 11, 2011 at 8:26 AM, Malte Brill revolut...@derbrill.de wrote:

 Ok,

 I will not claim I understand how this works, but it appears to do. :-)
 Still needs more testing, but for now I think it works.

 Cheers,

 Malte

 I have been asking a colleague for help (Danke danke danke Steffen) and
 here is what he came up with:

 local sRelations, sOpList

 on mouseUp
local tValid
delete variable sRelations
delete variable sOpList
set the itemdel to TAB
-- fld Data is TAb delimited. Person1 TAB Operator TAB Person2
repeat for each line theLine in fld data
if item 2 of theLine = = then
put = into sRelations[item 1 of theLine][item 3 of theLine]
put = into sRelations[item 3 of theLine][item 1 of theLine]
end if
if item 2 of theLine =  then
put  into sRelations[item 1 of theLine][item 3 of theLine]
put  into sRelations[item 3 of theLine][item 1 of theLine]
end if
if item 2 of theLine =  then
put  into sRelations[item 1 of theLine][item 3 of theLine]
put  into sRelations[item 3 of theLine][item 1 of theLine]
end if
end repeat
answer checkForValidRelation(the text of fld pers1,the text of fld
 pers2,the label of btn relation)
-- returns true or false for the validity of the relation you are about
 to set.
 end mouseUp

 private function checkForValidRelation pPers1,pPers2,pRelation
   local tResolved
   put resolveRelation(pPers1,pPers2,pRelation) into tResolved
   if not tResolved then return true
   return (checkOperator(pRelation) = pRelation)
 end checkForValidRelation

 private function resolveRelation pPers1,pPers2,pRelation
if sRelations[pPers1][pPers2] is not empty then
put sRelations[pPers1][pPers2]  cr after sOpList
return true
end if
-- iterate through list
return traverseList(pPers1,pPers2,pRelation)
 end resolveRelation

 private function traverseList pPers1,pPers2,pRelation
local tOther
put the keys of sRelations[pPers1] into tOther
repeat for each line personB in tOther
if personB = pPers2 then
put sRelations[pPers1][personB]  cr after sOpList
return true
else
if traverseList(pPers1,personB,pRelation) then
return true
 end if
end if
end repeat
 return false
 end traverseList

 private function checkOperator pRelation
local tSame, tLast
put true into tSame
put pRelation into tLast
repeat for each line theLine in sOpList
if theLine is not empty then
if theLine  pRelation then
put false into tSame
end if
put theLine into tLast
end if
end repeat
if not tSame then
return tLast
end if
return pRelation
 end checkOperator



 ___
 use-livecode mailing list
 use-livecode@lists.runrev.com
 Please visit this url to subscribe, unsubscribe and manage your
 subscription preferences:
 http://lists.runrev.com/mailman/listinfo/use-livecode

___
use-livecode mailing list
use-livecode@lists.runrev.com
Please 

Re: Coding Challenge

[Nonsanity]

On Fri, Mar 11, 2011 at 1:54 PM, Nonsanity f...@nonsanity.com wrote:

 Try this. I haven't tested, but the logic looks sound enough...


Actually, don't try that, try this. That one had flaws this one should fix.
(But again, not tested at ALL.)



private function resolveRelation pPers1,pPers2,pRelation
   put sRelations[pPers1][pPers2] into relationship

   -- if the relationship is a direct match, return true
   if relationship = pRelation then
  return true
   -- else if there is a direct relationship, but not the one we're testing
for, then fail
   else if relationship is not empty and relationship is not pRelation then
  return false
   end if

   -- make a reversed copy of our target relationship
   put pRelation into Rel
   if pRelation is  then put  into Rel
   else if pRelation is  then put  into Rel

   -- otherwise, there is no direct relationship, so look for an indirect
one
   -- to find one, we first find other items related to both inputs that
have the same relationship exactly
   -- then iterate through the tree looking for further = or =
relationships
   local tOther
   put the keys of sRelations[pPers1] into tOther
   repeat for each line personB in tOther
  if sRelations[pPers1][personB] = pRelation then
 get traverseList(pPers1,personB,pRelation,Rel)
 if it is true then
set sRelations[pPers1][pPers2] to pRelation
set sRelations[pPers2][pPers1] to Rel
return true
 end if
  end if
   end repeat

   -- test for the other direction, reverse everything
   put the keys of sRelations[pPers2] into tOther
   repeat for each line personB in tOther
  if sRelations[pPers2] = Rel then
 get traverseList(pPers2,personB,Rel,Rel)
 if it is true then
set sRelations[pPers1][pPers2] to pRelation
set sRelations[pPers2][pPers1] to Rel
return true
 end if
  end if
   end repeat

   return false
end resolveRelation


-- pPersV varies as we move through the tree
-- pPersX is our target person
private function traverseList pPersV,pPersX,pRelation,pRevRel
   -- are the two people of the correct relationship and/or equal?
   if sRelations[pPersV][pPersX] = pRelation or sRelations[pPersV][pPersX] =
= then
  return true
   end if
   local tOther
   put the keys of sRelations[pPersV] into tOther
   repeat for each line personB in tOther
  -- only follow branches leading to the same relationship we are
testing (= or =)
  -- So if: AB and BC and C=D and DE then AE must be true
  if sRelations[pPersV][personB] = pRelation or
sRelations[pPersV][personB] = = then
 if traverseList(personB,pPersX,pRelation) then
-- save this data directly for future speed
set sRelations[pPersV][pPersX] to pRelation
set sRelations[pPersX][pPersV] to pRevRel
return true
 end if
  end if
   end repeat
   return false
end traverseList


 ~ Chris Innanen
 ~ Nonsanity
___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding Challenge

[Nonsanity]

On Fri, Mar 11, 2011 at 2:32 PM, Nonsanity f...@nonsanity.com wrote:

 On Fri, Mar 11, 2011 at 1:54 PM, Nonsanity f...@nonsanity.com wrote:

 Try this. I haven't tested, but the logic looks sound enough...


 Actually, don't try that, try this. That one had flaws this one should fix.
 (But again, not tested at ALL.)



Typo.  Replace Rel,Rel with Rel,pRelation in the previous code.


 ~ Chris Innanen
 ~ Nonsanity
___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding Challenge

[Alex Tweedly]

You could use getRelationship() as I sent earlier, but it would be a bit 
clumsy. Easier to test for validity directly, something like (off the 
top of my head ...)


function canItBeValid p1, p2, pRelationship
  put sample(p1) into p1
  put sample(p2) into p2
  switch pRelationship
  case =
 if p1 is among the lines of gOlder[p2] then return false
 if p2 is among the lines of gOlder[p1] then return false
 return true
  case 
 if p1 = p2 then return false
 if p1 is among the lines of gOlder[p2] then return false
 return true
  case 
 if p1 = p2 then return false
 if p2 is among the lines of gOlder[p1] then return false
 return true
   default
  ask bad relationship  pRelationship
  return false
  end switch
end canItBeValid


but I see from a later email you have something already working, so go 
with that :-)


-- Alex.


On 11/03/2011 11:48, Malte Brill wrote:

Hi Alex,

well what I need to do is to check if the relationship I am wanting to set is 
valid or not, before I write it to the database. I might not be clear enough. 
But my goal is not really to check for existing relations, but the validity of 
the data I am about to enter into the Database. So if there is no relation at 
all, any relation is valid. Direct relations are clear and indirect relations 
need to be checked for.

Maybe I just dot understand how I would use your script in that case.

Sorry for the confusion.

Malte


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode




___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding Challenge

[Alex Tweedly]

Hi Malte. Hope I'm not too late to the party :-)

Here's some sample code that appears to work on my small sample data. 
It's not easy to come up with good test data here - nothing obvious 
could generate realistic test data by program.


Summary is

1. deal with all equality cases first. For every set of equal-age names, 
we use the same name (first one alphabetically) for the 'younger' 
relationship table.


2. normalize the realtionships (always younger, replace by same-age 
equivalent


3. for each name, keep two lists - those older and still to be handled, 
and those older but already dealt with


4. repeatedly, take a name from the first list and try to move it to the 
second. If this name itself has nothing left on its 'still-to-do' list 
then we have completely moved this name - remove it from the still-to-do 
list.


This should be reasonably efficient, though there are possible 
optimizations by keeping back-pointers - so let me know if this is too 
slow when you start using thousands of names,


-- Alex.


global gSameAge   -- array of those who are the same age
global gRelate -- array of direct (i.e. defined) relationships
global gOlder -- array containing (one per line) all those who are 
known older than the key

global gOlderStillToDo

on c
   put URL file:~/data.txt into tData

   put empty into gSameAge
   put empty into gOlder
   put empty into gOlderStillToDo
   -- deal with all the equal age info
   -- for each group of same-age people, we will use the first 
alphabetically

   -- step 1. normalize each individual equality to be alphabetocal
   -- step 2. sort the set of equality facts to be alphabetical
   -- step 3. for each subsequent name, store the earliest equal one
   repeat for each line  L in tData
  if word 2 of L = = then
 if word 1 of L  word 3 of L then
put word 1 of L  word 3 of L  CR after tEqual
 else
put word 3 of L  word 1 of L  CR after tEqual
 end if
  end if
   end repeat
   -- sort these to handle cases like a=b and b=d
   sort lines of tEqual
   repeat for each line L in tEqual
  if gSameAge[word 1 of L] is not empty then
 put gSameAge[word 1 of L] into gSameAge[word 2 of L]
  else
 put word 1 of L into gSameAge[word 2 of L]
  end if
   end repeat

   -- now normalize the direct relationships
   put empty into gRelate
   repeat for each line  L in tData
  if word 2 of L =  then
 put sample(word 1 of L) into t1
 put sample(word 3 of L) into t2
 put t1  t2  CR after tUnequal
  end if
   end repeat

   -- first the direct relationships
   repeat for each line L in tUnequal
  put word 2 of L  CR after gOlderStillToDo[word 1 of L]
   end repeat

   repeat 1 timeS -- forever, but avoid infinite loops
  put 0 into tNeedToDoAgain
  put the keys of gOlderStillToDo into tKeys
  -- for each one that still has something to do
  repeat for each line K in tKeys
 put empty into tNewlyFound
 put empty into tStillToDo
 repeat for each line L in gOlderStillToDo[K]
if L is not among the keys of gOlderStillToDo then
   -- add this older one, *and* all names known to be 
older than it

   put L  CR after gOlder[K]
   put gOlder[L]  CR after tNewlyFound
else -- this one hasn't yet been done, so we may need 
another iteration

   put L  CR after tStillToDo
   add 1 to tNeedToDoAgain
end if
 end repeat
 if tStillToDo is empty then
delete variable gOlderStillToDo[K]
 else
put tStillToDo into gOlderStillToDo[K]
 end if
 repeat for each line L in tNewlyFound
if L = K then   -- there is a loop in age info - i.e. bad 
input !!

   ask loop  L  K
   exit to top
end if
if L is not among the lines gOlder[K] and L is not among 
the lines of gOlderStillToDo[K] then

   put L  CR after gOlderStillToDo[K]
   add 1 to tNeedToDoAgain
end if
 end repeat
  end repeat
  if tNeedToDoAgain = 0 then exit repeat
   end repeat
   repeat for each key K in gOlder
  put K  ---  CR after field F
  put gOlder[K] after field F

   end repeat
end c

function sample pA
   if gSameAge[pA] is not empty then
  return gSameAge[pA]
   else
  return pA
   end if
end sample





On 08/03/2011 18:49, Malte Brill wrote:

Thanks for the head ups folks,

Björnke and Mark: I do not have the exact ages to sort by. All I do have is 
relations:

Malte is older than Björnke
Mark is older than Malte

And now I need to compute if it is valid to say:
- Björnke is older than Mark  (obviously not)
- Björnke is the same age as Mark (obviously not)
- Björnke is younger than Mark. (That´s the one)

What comes easy to the human brain in fact appears to be a lot more difficult 
when having to be 

Re: Coding Challenge

[Björnke von Gierke]

I'm not sure i understand the problem. given:

Paul 3
Peter 6
Joanne 6
fritz 9
Madelaine 15

sort by word 2 of each
put lineoffset(paul, theList) - lineoffset (peter) into theCompare
if theCompare   0 then
--child one is younger
else
--child 2 is younger
end if

that should solve all cases, unless they're the same age, which you need to 
check with a direct comparision (if word 2 of child1 = word 2 of child2 then 
sameAge)...

of course you could as well always use comparisions, and i'm sure you'd think 
of that yourself. so i really don't understand the question it seems.


On 8 Mar 2011, at 14:15, Malte Brill wrote:

 Hi folks,
 
 this is over my head. I have a problem I can not tackle alone, so please let 
 me scream for help...
 
 
 HEELLLP
 
 
 Imagine a group of kids:
 
 Paul,Peter,Fritz,Madeleine,Joanne (and any additional number of kids)
 
 I want to programmatically test on age relations of those kids:
 
 e.g. Peter is older than Paul, Joanne is the same age as Paul.
 Now I want to be able to test if the statement Joanne is older than Peter (or 
 same age, or younger) are valid or not. 
 I will need to be able to do this for all kids in the group. Any ideas on how 
 to tackle that???
 
 Cheers,
 
 Malte
 ___
 use-livecode mailing list
 use-livecode@lists.runrev.com
 Please visit this url to subscribe, unsubscribe and manage your subscription 
 preferences:
 http://lists.runrev.com/mailman/listinfo/use-livecode


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding Challenge

[Nonsanity]

Malte, are you looking for a general solution to problems such as this:

Tom is younger than Rose, but older than Will and Jack, in that order. Rose
is younger than Susie, but older than Jack. Jack is younger than Jim. Susie
is older than Rose, but younger than Jim. Jim is older than Tom. Who is the
oldest?

 ~ Chris Innanen
 ~ Nonsanity


On Tue, Mar 8, 2011 at 8:15 AM, Malte Brill revolut...@derbrill.de wrote:

 Hi folks,

 this is over my head. I have a problem I can not tackle alone, so please
 let me scream for help...


 HEELLLP


 Imagine a group of kids:

 Paul,Peter,Fritz,Madeleine,Joanne (and any additional number of kids)

 I want to programmatically test on age relations of those kids:

 e.g. Peter is older than Paul, Joanne is the same age as Paul.
 Now I want to be able to test if the statement Joanne is older than Peter
 (or same age, or younger) are valid or not.
 I will need to be able to do this for all kids in the group. Any ideas on
 how to tackle that???

 Cheers,

 Malte
 ___
 use-livecode mailing list
 use-livecode@lists.runrev.com
 Please visit this url to subscribe, unsubscribe and manage your
 subscription preferences:
 http://lists.runrev.com/mailman/listinfo/use-livecode

___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding Challenge

[Nonsanity]

On Tue, Mar 8, 2011 at 10:07 AM, Nonsanity f...@nonsanity.com wrote:

 Malte, are you looking for a general solution to problems such as this:

 Tom is younger than Rose, but older than Will and Jack, in that order.
 Rose is younger than Susie, but older than Jack. Jack is younger than Jim.
 Susie is older than Rose, but younger than Jim. Jim is older than Tom. Who
 is the oldest?

  ~ Chris Innanen
  ~ Nonsanity


Using that as an example, the steps to get not just the oldest but a
complete ordered list is to:

1) Codify all the less-than greater-than pairs and group them in a list all
going the same way. I'll pick less-than here. (Names have been reduced to
the first letter, except for Jim who is M.)

T  R
W  T
J  T
J  R
W  R
R  S
J  R
J  M
R  S
S  M
T  M

2) IF you have enough relationships to determine a complete ordering, then
the names missing from each side of the list represent the youngest and
oldest. In this case, M is not on the left (younger) side, and J is not
on the right (older) side, so Jim is the youngest and Jack is the oldest.

3) Eliminate all pairings that involve the known individuals and see if you
have enough information to proceed.

(youngest) J ... M (oldest)

T  R
W  T
   [J  T]
   [J  R]
W  R
R  S
   [J  R]
   [J  M]
R  S
   [S  M]
   [T  M]

4) Repeat. S is not on the left (younger) side and W is not on the right
(older) side, so add them to the ordered list.

(youngest) J W ... S M (oldest)

T  R

5) With only two names remaining, and one conditional, we have enough
information to complete the list.

(youngest) J W T R S M (oldest)


If the data given is not enough to get a complete list, it may be possible
to pluck out some slightly more complex relationships (A is older than B IF
C is older than D) but it sounds like you need more solid answers.

Anyway, that's A general solution to this sort of problem. I don't know if
there's a better one or not... I didn't already know this proceedure, I just
worked it out when you asked. I only now went back to the web site where I
got the sample problem and clicked the answer button. It's Jim. Yeay.

Hmmm... Looking back, I think that if you have incomplete data, you can
still get a mostly ordered list out of this. If more than one name is
missing from the left (younger) side, then you know that those names are the
youngest, you just don't know which is younger or older among them. I'd
repackage them as a single unit (new letter) and substitute it though the
whole list for all the affected names, then try again. In the end, you
should have a list with groups of unknown order scattered throughout, but
still have an overall list. If further data is presented for the names in
those groups, they can be put through this same routine in isolation from
the rest of the list.

Hope this is what you were looking for! (But it was fun to work out
anyway...)


 ~ Chris Innanen
 ~ Nonsanity
___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding Challenge

[Nonsanity]

Oh!

I forgot to add J  W to the list when I re-did it in the email. It comes
from the in that order phrase below. Without it, there are TWO names
missing from the youngest side of the list, so W and J would be grouped
together (into X) like so:

T  R
X  T
X  T
X  R
X  R
R  S
X  R
X  M
R  S
S  M
T  M

And the output of the first iteration would be:

(youngest) [J,W] ... M

The rest of the solution would proceed as normal. You would end up knowing
that Jack and Will are younger than everyone else, but not know which of the
two was the youngest.

Oh, and any equality statements in the original data (Karen is the same age
as Helen) should be substituted for a single variable, grouping them the
same way you would a pair of unknowns, like J and W when I left out the one
conditional statement.


 ~ Chris Innanen
 ~ Nonsanity



On Tue, Mar 8, 2011 at 10:41 AM, Nonsanity f...@nonsanity.com wrote:

 On Tue, Mar 8, 2011 at 10:07 AM, Nonsanity f...@nonsanity.com wrote:

 Malte, are you looking for a general solution to problems such as this:

 Tom is younger than Rose, but older than Will and Jack, in that order.
 Rose is younger than Susie, but older than Jack. Jack is younger than Jim.
 Susie is older than Rose, but younger than Jim. Jim is older than Tom. Who
 is the oldest?

  ~ Chris Innanen
  ~ Nonsanity


 Using that as an example, the steps to get not just the oldest but a
 complete ordered list is to:

 1) Codify all the less-than greater-than pairs and group them in a list all
 going the same way. I'll pick less-than here. (Names have been reduced to
 the first letter, except for Jim who is M.)

 T  R
 W  T
 J  T
 J  R
 W  R
 R  S
 J  R
 J  M
 R  S
 S  M
 T  M

 2) IF you have enough relationships to determine a complete ordering, then
 the names missing from each side of the list represent the youngest and
 oldest. In this case, M is not on the left (younger) side, and J is not
 on the right (older) side, so Jim is the youngest and Jack is the oldest.

 3) Eliminate all pairings that involve the known individuals and see if you
 have enough information to proceed.

 (youngest) J ... M (oldest)

 T  R
 W  T
[J  T]
[J  R]
 W  R
 R  S
[J  R]
[J  M]
 R  S
[S  M]
[T  M]

 4) Repeat. S is not on the left (younger) side and W is not on the
 right (older) side, so add them to the ordered list.

 (youngest) J W ... S M (oldest)

 T  R

 5) With only two names remaining, and one conditional, we have enough
 information to complete the list.

 (youngest) J W T R S M (oldest)


 If the data given is not enough to get a complete list, it may be possible
 to pluck out some slightly more complex relationships (A is older than B IF
 C is older than D) but it sounds like you need more solid answers.

 Anyway, that's A general solution to this sort of problem. I don't know if
 there's a better one or not... I didn't already know this proceedure, I just
 worked it out when you asked. I only now went back to the web site where I
 got the sample problem and clicked the answer button. It's Jim. Yeay.

 Hmmm... Looking back, I think that if you have incomplete data, you can
 still get a mostly ordered list out of this. If more than one name is
 missing from the left (younger) side, then you know that those names are the
 youngest, you just don't know which is younger or older among them. I'd
 repackage them as a single unit (new letter) and substitute it though the
 whole list for all the affected names, then try again. In the end, you
 should have a list with groups of unknown order scattered throughout, but
 still have an overall list. If further data is presented for the names in
 those groups, they can be put through this same routine in isolation from
 the rest of the list.

 Hope this is what you were looking for! (But it was fun to work out
 anyway...)


  ~ Chris Innanen
  ~ Nonsanity


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding Challenge

[Nonsanity]

On Tue, Mar 8, 2011 at 10:59 AM, Nonsanity f...@nonsanity.com wrote:

 Without it, there are TWO names missing from the youngest side of the
 list...


Correction:
...from the OLDEST side of the list, and therefore are the youngest...

Gotta keep things clear in stuff like this. I almost let that slide, but...
:)


 ~ Chris Innanen
 ~ Nonsanity
___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding Challenge

[Malte Brill]

Thanks for the head ups folks,

Björnke and Mark: I do not have the exact ages to sort by. All I do have is 
relations:

Malte is older than Björnke
Mark is older than Malte

And now I need to compute if it is valid to say: 
- Björnke is older than Mark  (obviously not)
- Björnke is the same age as Mark (obviously not)
- Björnke is younger than Mark. (That´s the one)

What comes easy to the human brain in fact appears to be a lot more difficult 
when having to be tackled computationaly.

Chris: Yes, I want such a list in the end. But in order to finally get this I 
will need to tell the machine which relations are legal first (the user tells 
which relations there are) and ideally filter out the data for relations that 
make no sense. Now I wish it was easy to tell the machine to just use logic and 
make sense itself :D This comparison has to be done for thousands of entities 
(children in this case). 

Cheers,

Malte


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding Challenge

[Peter Haworth]

Would it work to assign pseudo ages to each of the people based on the given 
relationships, starting with the first?  In your example:

Malte (1000) is older than Bjornke(500)
Mark (1500) is older than Malte

Then you can sort by the age as others have suggested.  Seems to work well on 
this small sample size but I guess it could get pretty complicated to assign 
the pseudo ages if there is a large number of people.

Pete Haworth

On Mar 8, 2011, at 10:49 AM, Malte Brill wrote:

 Thanks for the head ups folks,
 
 Björnke and Mark: I do not have the exact ages to sort by. All I do have is 
 relations:
 
 Malte is older than Björnke
 Mark is older than Malte
 
 And now I need to compute if it is valid to say: 
 - Björnke is older than Mark  (obviously not)
 - Björnke is the same age as Mark (obviously not)
 - Björnke is younger than Mark. (That´s the one)
 
 What comes easy to the human brain in fact appears to be a lot more difficult 
 when having to be tackled computationaly.
 
 Chris: Yes, I want such a list in the end. But in order to finally get this I 
 will need to tell the machine which relations are legal first (the user tells 
 which relations there are) and ideally filter out the data for relations that 
 make no sense. Now I wish it was easy to tell the machine to just use logic 
 and make sense itself :D This comparison has to be done for thousands of 
 entities (children in this case). 
 
 Cheers,
 
 Malte
 
 
 ___
 use-livecode mailing list
 use-livecode@lists.runrev.com
 Please visit this url to subscribe, unsubscribe and manage your subscription 
 preferences:
 http://lists.runrev.com/mailman/listinfo/use-livecode
 


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding Challenge

[Bob Sneidar]

Dang I was just formulating that, but you would have to build a list of already 
assigned people and then pop them in the middle as logic dictates. 

Bob


On Mar 8, 2011, at 11:05 AM, Peter Haworth wrote:

 Would it work to assign pseudo ages to each of the people based on the 
 given relationships, starting with the first?  In your example:
 
 Malte (1000) is older than Bjornke(500)
 Mark (1500) is older than Malte
 
 Then you can sort by the age as others have suggested.  Seems to work well on 
 this small sample size but I guess it could get pretty complicated to assign 
 the pseudo ages if there is a large number of people.
 
 Pete Haworth
 
 On Mar 8, 2011, at 10:49 AM, Malte Brill wrote:
 
 Thanks for the head ups folks,
 
 Björnke and Mark: I do not have the exact ages to sort by. All I do have is 
 relations:
 
 Malte is older than Björnke
 Mark is older than Malte
 
 And now I need to compute if it is valid to say: 
 - Björnke is older than Mark  (obviously not)
 - Björnke is the same age as Mark (obviously not)
 - Björnke is younger than Mark. (That´s the one)
 
 What comes easy to the human brain in fact appears to be a lot more 
 difficult when having to be tackled computationaly.
 
 Chris: Yes, I want such a list in the end. But in order to finally get this 
 I will need to tell the machine which relations are legal first (the user 
 tells which relations there are) and ideally filter out the data for 
 relations that make no sense. Now I wish it was easy to tell the machine to 
 just use logic and make sense itself :D This comparison has to be done for 
 thousands of entities (children in this case). 
 
 Cheers,
 
 Malte
 
 
 ___
 use-livecode mailing list
 use-livecode@lists.runrev.com
 Please visit this url to subscribe, unsubscribe and manage your subscription 
 preferences:
 http://lists.runrev.com/mailman/listinfo/use-livecode
 
 
 
 ___
 use-livecode mailing list
 use-livecode@lists.runrev.com
 Please visit this url to subscribe, unsubscribe and manage your subscription 
 preferences:
 http://lists.runrev.com/mailman/listinfo/use-livecode


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding Challenge

[Peter Haworth]

I've got a feeling that any mathematician lurking on this list could give the 
algorithm to do this in no time - something to do with set theory?.  My 
clumsy attempt at doing would probably involve creating an array with the 
person's name as the key and the assigned pseudo age as the value.  Then I'd 
probably make multiple passes through the list assigning the ages based on the 
relationship and whether there is a key in the array for the person.  After 
that, it's easy to compare:

if Array[Bjornke]  Array[Malte] then
do true stuff
else
do false stuff
end if  

Pete Haworth

On Mar 8, 2011, at 11:14 AM, Bob Sneidar wrote:

 Dang I was just formulating that, but you would have to build a list of 
 already assigned people and then pop them in the middle as logic dictates. 
 
 Bob
 
 
 On Mar 8, 2011, at 11:05 AM, Peter Haworth wrote:
 
 Would it work to assign pseudo ages to each of the people based on the 
 given relationships, starting with the first?  In your example:
 
 Malte (1000) is older than Bjornke(500)
 Mark (1500) is older than Malte
 
 Then you can sort by the age as others have suggested.  Seems to work well 
 on this small sample size but I guess it could get pretty complicated to 
 assign the pseudo ages if there is a large number of people.
 
 Pete Haworth
 
 On Mar 8, 2011, at 10:49 AM, Malte Brill wrote:
 
 Thanks for the head ups folks,
 
 Björnke and Mark: I do not have the exact ages to sort by. All I do have is 
 relations:
 
 Malte is older than Björnke
 Mark is older than Malte
 
 And now I need to compute if it is valid to say: 
 - Björnke is older than Mark  (obviously not)
 - Björnke is the same age as Mark (obviously not)
 - Björnke is younger than Mark. (That´s the one)
 
 What comes easy to the human brain in fact appears to be a lot more 
 difficult when having to be tackled computationaly.
 
 Chris: Yes, I want such a list in the end. But in order to finally get this 
 I will need to tell the machine which relations are legal first (the user 
 tells which relations there are) and ideally filter out the data for 
 relations that make no sense. Now I wish it was easy to tell the machine to 
 just use logic and make sense itself :D This comparison has to be done for 
 thousands of entities (children in this case). 
 
 Cheers,
 
 Malte
 
 
 ___
 use-livecode mailing list
 use-livecode@lists.runrev.com
 Please visit this url to subscribe, unsubscribe and manage your 
 subscription preferences:
 http://lists.runrev.com/mailman/listinfo/use-livecode
 
 
 
 ___
 use-livecode mailing list
 use-livecode@lists.runrev.com
 Please visit this url to subscribe, unsubscribe and manage your subscription 
 preferences:
 http://lists.runrev.com/mailman/listinfo/use-livecode
 
 
 ___
 use-livecode mailing list
 use-livecode@lists.runrev.com
 Please visit this url to subscribe, unsubscribe and manage your subscription 
 preferences:
 http://lists.runrev.com/mailman/listinfo/use-livecode
 


___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding Challenge

[Brian Yennie]

Malte,

How about something like this:

1) Normalize all of your relations to older than (reverse of younger than)

So if you started with:

Malte is older than Bjornke
Mark is older than Malte
Mark is younger than Fred
George is younger than Fred

You now have these ordered pairs:

Malte,Bjornke
Mark, Malte
Fred,Mark
Fred, George

Ignore equals relations for now.

2) Combine any places where a name occurs in both the 1st and last items, 
repeat as necessary:

=
Malte,Bjornke
Fred, Mark, Malte
Fred, George

=
Fred, Mark, Malte, Bjornke
Fred, George

4) For any equals relations, simply treat them as aliases in the final 
analysis. If a name doesn't appear in your list, try all of the aliases because 
the result will be the same.

5) To answer a question, use each resulting line as a test.

Examples: 
If Person A  Person B, then they will appear before them in at least 1 line
If Person A  Person B, then they will appear after them in at least 1 line
If Person A = Person B, then they must not appear on the same line

This probably isn't perfect, but maybe a good direction?


 Thanks for the head ups folks,
 
 Björnke and Mark: I do not have the exact ages to sort by. All I do have is 
 relations:
 
 Malte is older than Björnke
 Mark is older than Malte
 
 And now I need to compute if it is valid to say: 
 - Björnke is older than Mark  (obviously not)
 - Björnke is the same age as Mark (obviously not)
 - Björnke is younger than Mark. (That´s the one)
 
 What comes easy to the human brain in fact appears to be a lot more difficult 
 when having to be tackled computationaly.
 
 Chris: Yes, I want such a list in the end. But in order to finally get this I 
 will need to tell the machine which relations are legal first (the user tells 
 which relations there are) and ideally filter out the data for relations that 
 make no sense. Now I wish it was easy to tell the machine to just use logic 
 and make sense itself :D This comparison has to be done for thousands of 
 entities (children in this case). 
 
 Cheers,
 
 Malte
___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding Challenge

[Nonsanity]

That's a lot of children. Hmmm... Are you looking to use relative ages
instead of actual ages because the entry times (old photographs for example)
involve the children at different ages? The entry of all that is going to be
the biggest headache, I imagine, as every relationship will require
selecting two people.

But once the data is input, the method I outlined should give you the
results you want.

 ~ Chris Innanen
 ~ Nonsanity


On Tue, Mar 8, 2011 at 1:49 PM, Malte Brill revolut...@derbrill.de wrote:

 Thanks for the head ups folks,

 Björnke and Mark: I do not have the exact ages to sort by. All I do have is
 relations:

 Malte is older than Björnke
 Mark is older than Malte

 And now I need to compute if it is valid to say:
 - Björnke is older than Mark  (obviously not)
 - Björnke is the same age as Mark (obviously not)
 - Björnke is younger than Mark. (That´s the one)

 What comes easy to the human brain in fact appears to be a lot more
 difficult when having to be tackled computationaly.

 Chris: Yes, I want such a list in the end. But in order to finally get this
 I will need to tell the machine which relations are legal first (the user
 tells which relations there are) and ideally filter out the data for
 relations that make no sense. Now I wish it was easy to tell the machine to
 just use logic and make sense itself :D This comparison has to be done for
 thousands of entities (children in this case).

 Cheers,

 Malte


 ___
 use-livecode mailing list
 use-livecode@lists.runrev.com
 Please visit this url to subscribe, unsubscribe and manage your
 subscription preferences:
 http://lists.runrev.com/mailman/listinfo/use-livecode

___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Re: Coding Challenge

[JosepM]

And if you assign a value (a weight) to the relation? 0 or 1 in function if
are older or younger, assign the relation value to the person and compare
the persons with the relation value or weigth.

Salut,
Josep

--
View this message in context: 
http://runtime-revolution.278305.n4.nabble.com/Coding-Challenge-tp3341313p3342144.html
Sent from the Revolution - User mailing list archive at Nabble.com.

___
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode