Re: Ranges

[Gaute B Strokkenes <biggaute@uwc.net>]

On 5 jan 2004, larsi@gnus.org wrote:

> I've been thinking about the range thing -- using ranges instead of
> lists whenever talking about articles.  And I've got some ideas
> cooking about the C implementation of those, and I think it could
> possibly be quite groovy.  (A new data type; super-hyper-fast
> `range-memq', etc.)
>
> But, on the other hand -- what's the impact, really?

The best (in my not-so-humble opinion) argument, and one that hasn't
been mentioned so far, is that consistently using ranges means that
you only use one data structure to indicate sets of articles.  I have
been confused on more than one ocassion while browsing through the
gnus source whether ranges or lists are expected.

> Where does using lists instead of ranges really represent a
> noticeable performance drag?  Ok, when entering a group with 1000
> articles, you get a list that's 1000 elements long, but does
> creating that list really make any noticeable difference?
>
> I'd like to see some scenarios where this matters before going down
> this road.

-- 
Gaute Strokkenes                        http://www.srcf.ucam.org/~gs234/
I'm in ATLANTIC CITY riding in a comfortable ROLLING CHAIR...