OpenEuphoria

Re: new tables

• Posted by Derek Parnell <ddparnell at bigpond.com> Sep 07, 2002
• 367 views

Yell
----------------
cheers,
Derek Parnell
----- Original Message -----
From: "Jiri Babor" <jbabor at PARADISE.NET.NZ>
To: "EUforum" <EUforum at topica.com>
Subject: new tables


>
>
> Hi Everybody,
>
> This is mainly for the wise ones, who do not visit my pages at all - it's
hardly
> worth it anyway. I have just posted a new version of associative lists,
this
> time called, briefly and correctly, tables ;). The basic scheme is
different
> from the 'parallel sequences' I used originally. It can be summarised as
>
>     {val1, val2, ... valN, {key1, key2, ... keyN}}
>
> The speed is roughly the same as previously, except for 'appending' of new
> entries, which is slightly faster.
>
> The only real advantage of the new scheme is that it's more intuitive in
> situations, when the entry index is already known. Then the most common
> functions set() and fetch() that rely on relatively slow find() can be
replaced
> by much faster simple assignments:
>
>     table[i] = new_value
>
> and
>
>     value = table[i]
>
> respectively, where 'i' is obviously the previously determined index.
>
> This simple library is quite good for small to medium sized tables. If
your
> tables contain more than about 200 entries, you might be better off with
sorted
> tables. For those I use a binary search, or alternatively, a 'hybrid'
variant,
> which relies on a mixture of find() and binary search, depending on the
size of
> the particular table. Just give me a yell, if you are *really* interested.
>
> jiri
>
>
>
>