C
Charles Hoffman
I think this is just a common mistake among newbie-ish programmers.
Which isn't to say anything about Javier specifically, I don't know him,
but in general, a review of one's Data Structures course may be in order
when this happens. But I think it happens to just about every
programmer at some point. I remember the first time I tried to do some
ASP scripting for some web pages and I decided to use the Dictionary
object. At one point I wanted to loop through the Dictionary and
display a listing of its contents. I got all pissed off when I realized
that the loop didn't return the items in alphabetical order. Only later
did I realize that the underlying data structure is a hash, and that the
whole point of a hash is to maximize speed of access by sacrificing
things like, being able to iterate it in any particular order. Of
course, the word "Dictionary" does, for most of us, bring to mind
something that's in alphabetical order, so maybe it was Microsoft's
fault for givig it a dumb name. At any rate, it would be a bad idea to
allow what is essentially a mistake made by programmers to force an
abnormal, slower implementation onto a data structure whose definition
is largely agreed upon in computing. In other words, I'm with the "we
don't need an 'ordered hash'" crowd. I think that, besides slowing it
down, it goes against the very definition of what a hash is.
If you need to iterate through a hash in order by key, it's a simple
matter to grab an array of the keys first, sort it, then iterate from
that. I'm probably going to embarrass myself with my Ruby nuby-ness
here, but off the top of my head:
my_hash.keys.sort.each{|key| doSomethingWith(my_hash[key])}
If you want, and if you really must do away with the overhead of having
to retrieve the key array and sort it first, you could modify the Hash
class with in your own program (Ruby lets you do stuff like that) and
have it keep a sorted array of keys, by adding each new key to it in a
sorted position, after every time an item is added to the hash, and
splicing the key out of the array whenever an item is removed. Then you
could override the each method, and the inspect method if necessary.
--ch--
Which isn't to say anything about Javier specifically, I don't know him,
but in general, a review of one's Data Structures course may be in order
when this happens. But I think it happens to just about every
programmer at some point. I remember the first time I tried to do some
ASP scripting for some web pages and I decided to use the Dictionary
object. At one point I wanted to loop through the Dictionary and
display a listing of its contents. I got all pissed off when I realized
that the loop didn't return the items in alphabetical order. Only later
did I realize that the underlying data structure is a hash, and that the
whole point of a hash is to maximize speed of access by sacrificing
things like, being able to iterate it in any particular order. Of
course, the word "Dictionary" does, for most of us, bring to mind
something that's in alphabetical order, so maybe it was Microsoft's
fault for givig it a dumb name. At any rate, it would be a bad idea to
allow what is essentially a mistake made by programmers to force an
abnormal, slower implementation onto a data structure whose definition
is largely agreed upon in computing. In other words, I'm with the "we
don't need an 'ordered hash'" crowd. I think that, besides slowing it
down, it goes against the very definition of what a hash is.
If you need to iterate through a hash in order by key, it's a simple
matter to grab an array of the keys first, sort it, then iterate from
that. I'm probably going to embarrass myself with my Ruby nuby-ness
here, but off the top of my head:
my_hash.keys.sort.each{|key| doSomethingWith(my_hash[key])}
If you want, and if you really must do away with the overhead of having
to retrieve the key array and sort it first, you could modify the Hash
class with in your own program (Ruby lets you do stuff like that) and
have it keep a sorted array of keys, by adding each new key to it in a
sorted position, after every time an item is added to the hash, and
splicing the key out of the array whenever an item is removed. Then you
could override the each method, and the inspect method if necessary.
--ch--
