World: r3wp

I mean - after seeing Don's presentation ....

hehe, well Ron is 15km from work  ;-)

so, I am priviledged   :-)

carl also seems to like liquid  :-D  and liked liquidGL a lot...

so there might be a way to get relavance and RT as the engine, and 
elixir as the desktop  :-)

liquidGL.. I need to roam more rebol sites..

it would allow millions of nodes

his system is not specifically meant for dataflow "processing" but 
can be used to support its specification, so we could actually store 
the data into relavance, I guess.

i need more sugar...

BTW, my demo was supposed to include a networked node system, but 
a snag in my hasty implimentation kept me from finishing in time 
for the devcon.

this allows us to send the node data inbetween two sessions of elixir 
.

you see, graham is right, you never finish anything.

:P

hehe

Mmmmm beem me up scotty..... oke and now my clothes....

well who said I was done ;-)

whould be done by next week, I guess I'll keep it for the actual 
file release.

500 million items "Starts" to slow down

(relavance that is!)

that's acceptable

:)

note that if we where to equate that to cell count we can easily 
multiply by 10, maybe much more.

how do you fit 500 Million in a Graphical presentaion ?

you filter out what you are working on.

well as a overall display its still 1 million...

1 dot per inch..

Who is on next?

could anyone take a high-res photo of the screen?

Q:  How many companies or organisations have implemented an associative 
database up til now ?

that'd be great, as it's very hard to recognize anything (apart from 
random windows) on the stream ...

Q: how quick is searching in the database...

Christian, I had a good look at their website when it was mentioned 
first here in rebol-land. I am pretty sure you can find the whole 
presentation there.

there is no searching.

but in a structure of 1 million how do i get from a -> z  quickly 
?

everything is indexed on entry.

every single item (cell)

They said the system requires modern CPUs because it takes more number 
crunching. (from memory about an order of magnitude more than relational 
databases.)

notice he sais "Find" not "Search"

oke thanks..

That's it. As always: Speed or Space.

one reason is that they kick out VM.

relavance does not use one single byte in VM, so its ram never gets 
paged to disk.

so you actually need more ram because of that... but actually, the 
nodes are all on disk.

so you can read and trash all nodes, at disk speed, so its very efficient 
actually.

anton, you mean relavance.com? can't find much there in terms of 
technical information ...

Can't be corrupted. I've got some ways that will help. But I guess 
he means the core system is very simple, small, and used very frequently, 
therefore hard to introduce bugs into.

Christian, I would follow the original blog article link, maybe.

IIRC they have different disk access methods, but have their own 
File system which kicks out the os, so that they can properly implement 
the DMA. also note that their memory model has no memory fragmentation 
 the available ram and disk space is always contiguous!!

I remember reading a fair bit of material similar to maybe the first 
third what we're seeing now.

so its not just the concepts which are interesting, its the actual 
implementation to.