World: r3wp

there are newer versions, but not published. the version of show 
for devcon 2005 was based on a newer version.

the one involving a (very) very big clock, I think is not a part 
of show.r :-)

Carl: "I would really like to have a DB built-into REBOL 3.0


The trick is, it cannot just be any DB... it has to align perfectly 
with REBOL. Specifically, we want the DB to be able to store all 
REBOL datatypes efficiently. But, we also want the DB code to be 
very small. I will not allow REBOL to bloat for it. And, finally, 
the code must be free, like SQLite is (at least, I think).


Note that the DB does not need to have every feature in the book. 
 I'm not asking for SQL (that is, we will use REBOL or a dialect 
to access the DB, no need for another language).


I would really like to get the help of the community in this effort, 
but I do not have time to even organize a decent search.  If someone 
else wants to head it up, it would very much improve R3.0. If not, 
then we will do RIF, which gets us a lot of DB, and is super efficient 
compared to any other DB. But, it is only an index system, not a 
RDB."

Please note that there is a very limited time to make a decision 
here. So if you have any idea / can help in any way, please speak. 
Otherwise, we're probably going to either just implement RIF or try 
to work something from scratch.

has berkley db been considered & dismissed?

what kind of license does it have (bsd?)

also, do you know about how small it can get (while still being useful)?

not sure just know it is used in many unix apps

will check

thanks.

ack  oracle bought sleepycat (maintainers if berkeley ) last month

not a good sign

http://www.sleepycat.com/products/bdb.html

# Support for memory constrained devices (footprint as small as 350KB)

Our open source license is OSI-certified and permits use of Berkeley 
DB in open source projects or in applications that are not distributed 
to third parties. Our commercial license permits closed-source distribution 
of an application to third parties and provides business assurance.

I'm not particularly advocating BDB, it is just the first low levle 
persistance layer that comes to mind.

I am only a lurker in database-things. But we have 1-2 sql-dialects? 
RebDb and somehthing from Coccinelle IIRC? I guess if these people 
get something super efficient in their hands, they could not resist 
to buidl an RDB n top of it?

http://www.sqlite.org/is worth looking at

RIF: REBOL Indexed Files
http://www.rebol.net/article/0007.html
http://www.rebol.net/article/0017.html

It was also mentioned at DevCon 2005.

Who's the guy who wants db built into Rebol? Guys - anything built 
directly into rebol is waste of space and will not please next guy! 
I expect Rebol 3 being a proper design - no inbuild stuff, please!

that kind of stuff really does not belong under the RT's radar. Give 
us general abstraction interfaces, e.g. RIF, not a concrete DB and 
if so, do it via some interface ... What is wrong with current sqlite 
driver? Well, maybe the .dll requirement which is not free ....

built in means always available. it means that everyone can assume 
it being present.

it's the same difference as having view or using whatever the os 
provides for the gui

I am not sure - even with rebol (core), if you want gui, you have 
to download View :-) I am not against kind of standardising upon 
good modules, but not sure if I want one big exe with everythign 
inside. :-)

Would you really want a non-native GUI, with every call to / from 
it going via slow [in relative terms] routine! APIs?


I agree that having one *huge* exe that does 100% of what *everyone* 
wants but only has 20% of it's functionality used by the average 
coder is a bad thing. The problem is defining what constitutes huge 
and what functionality is needed by most coders. Here are some typical 
components:

	Graphics
	Sound
	Networking
	Maths
	Encryption
	Data storage
	Installer
	Registry access
	Library access


Which should be built-in? Which should be loadable modules? Is your 
preference going to be the same as mine? These are not easy design 
questions.

What design question?

Every OS is done in modular way for a reason - give a break please 
- one of the best OS designs I looked into a bit imo is QNX - even 
something like event system for gui is external to kernel - in fact, 
the kernel is really small and nearly everything is module. Aren't 
we talking real-time OS here?

Besides that, IIRC Carl told us in the past, that currently what 
you find as a rebol component (look into system object), is already 
internally modular, so who knows - maybe it already goes via some 
abstraction api

What I am simply a bit scared about is missing the chance by doing 
the things right this time. I would really hate to see someone pushing 
Carl to include mp3 inside, while the oter guy will complain he/she 
wants ogg, and next one will require something from Microsoft (wmv?):-)

I have also suspicion, that some of you guys, e.g. Gabriele (I mean 
it in a friendly non complaining mode :-), don't mind having everything 
in one .exe. Neither do I in fact, View is still small in today's 
terms - but - if we talk small devices - it is not. Go and try to 
download it via GPRS I use with my cell phone - it does not look 
like a small app to upgrade in any way :-)

oh, and I don't believe in "dll hell" - I don't suffer any ... for 
me external component requriement (codec, math module, whatever) 
is just one other dependency .... your script can already have tonnes 
of other ones, so :-)

But it will be probably better to wait how  Carl describes "rebol 
like never before" in technical terms :-)

one thing is "everything in", another thing is basic functionality 
in.

if you want mp3 then you need to write a plugin for it. but we're 
talking about all the reblets that need to save their data somewhere

contact databases, chat programs, and millions of other reblets

you mean database/storage as a default? That makes sense ... but 
I thought (knowing little about what RIF actually was supposed to 
be), that it will be RIF .... e.g. for me, RebDB was good way - because 
of the form of representation - not that it was written in rebol 
- simply you could very easily mold any value, it was text readable 
and you could maintain it by other script easily ...

well, maybe I could regard its binary representation as a kind of 
rebin (whatever rebin was supposed to be :-) ....

rif is only rebol indexed file. an indexed file is not an rdb. of 
course, if no rdb matching our needs is found we'll just have rif...

a contact db, or chat program requires a RDBMS?

i'm guessing a flat-file would handle 99% of any reblets needs.

An ISAM-like solution is pretty handy for folks who want something 
more than plain-text storage but something less than a full-blown 
RDBMS.

I'd like a basic cross-platform data manager of the sort Ashley suggests.

That can then be used to build SQL type databases for those who want 
them. 

Or list type databases (for people like me who use that sort of approach 
a lot)
Or whatever.
No need to prematurely bind to the relational model/

What's your "list type databases"?

I'm all for "everything in" if it doesn't blow up the executable 
size. "Everything in" is one of the main reasons I use REBOL. Imagine 
if we had to include VID, network, etc. from external libraries...

An SQL type database is one where the basic unit of storage is an 
atomic data item, though packaged into rows / tuples.   Getting rid 
of 1NF data (recurring groups) is the first thing you are taught 
to do.

In a list type database the basic unit of storage is a list (think 
REBOL block). That's far from being 1NF, especially as a list can 
contain other lists.

.

.

.

Ok.. that's a bit excessive