New names for some operations taylanbayirli@xxxxxx (13 Sep 2015 12:13 UTC)
Re: New names for some operations John Cowan (13 Sep 2015 19:01 UTC)
Re: New names for some operations taylanbayirli@xxxxxx (13 Sep 2015 20:19 UTC)
Re: New names for some operations John Cowan (13 Sep 2015 22:20 UTC)
Re: New names for some operations taylanbayirli@xxxxxx (14 Sep 2015 08:29 UTC)
Re: New names for some operations Shiro Kawai (13 Sep 2015 23:28 UTC)
Re: New names for some operations John Cowan (13 Sep 2015 23:45 UTC)
Re: New names for some operations Shiro Kawai (13 Sep 2015 19:27 UTC)
Re: New names for some operations taylanbayirli@xxxxxx (13 Sep 2015 20:04 UTC)
Re: New names for some operations John Cowan (13 Sep 2015 20:11 UTC)
Re: New names for some operations Shiro Kawai (13 Sep 2015 20:25 UTC)
Re: New names for some operations taylanbayirli@xxxxxx (13 Sep 2015 21:46 UTC)
Re: New names for some operations taylanbayirli@xxxxxx (14 Sep 2015 09:39 UTC)
Re: New names for some operations Arthur A. Gleckler (13 Sep 2015 19:42 UTC)
Re: New names for some operations taylanbayirli@xxxxxx (13 Sep 2015 19:57 UTC)
Re: New names for some operations Arthur A. Gleckler (14 Sep 2015 03:25 UTC)
Re: New names for some operations John Cowan (14 Sep 2015 03:38 UTC)

Re: New names for some operations taylanbayirli@xxxxxx 13 Sep 2015 19:57 UTC

"Arthur A. Gleckler" <xxxxxx@speechcode.com> writes:

> On Sep 13, 2015 5:13 AM, "Taylan Ulrich Bayırlı/Kammer"
> <xxxxxx@gmail.com> wrote:
>
>> 2. specifying ordered hash tables like in PyPy.
>
> What exactly does "ordered hash table" mean?

Entries are ordered according to insertion order.

From what I've been told on #guile(*), it maps keys to indexes in a
regular hash table, and stores the real values in a vector.  That means
lookups involve an additional pointer dereference.  Deleted entries lead
to holes (wasted slots) in the values vector, but the table is compacted
when 50% or more allocated entries are dead.

(*) https://gnunet.org/bot/log/guile/2015-09-13#T750068

Taylan