This one was a stinker!
Line spaces as possibly spoilers below!!!
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
First mistake was assuming that directory names were unique. NO!
Next mistake - assuming that parent-child pairs were unique. NO!
I ended up with derived directory chained names, such as rt.a, rt.a.e
etc which is far
from elegant but did the trick - and took me hours to get my head around.
I keep an eye on the Q/K forum. Earlier postings there, but I imagine
they
used K's dictionary facility
. Perhaps boxed/nested arrays are the best approach
in J to representing the directory tree, but I didn't use them.
I do think it would be better if the day's example included the pitfalls
- yesterday's
example used unique names and parent-child pairs - but probably
deliberate?!
As it was, I had a reasonable, though rather verbose, function which
solved the
example, but whose answer to the day's data was wrong, and stayed
wrong through
various "improvements."
Mike
On 08/12/2022 15:05, David Lambert wrote:
I must misunderstand the instructions since I get same wrong answer in
every language I try.
Strategy: create a list of rooted path to all files. Make a set of
the directories-part, producing a unique list of all directories. For
each of these, add the file sizes for all files with matching leading
path. Sum the directory totals for those with less than or equal to
one hundred thousand. This strategy solves the example.
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
