World: r3wp

Note that if you view the tweets on Twitter itself, it links back 
to my page (where it says 'Client for REBOL') - http://twitter.com/rgrebol

So, what's missing?

And can your oauth work be used for google ?

Should be somewhat compatible.  There's two main parts: the header 
and the handshake.

I've tried to separate as much as my code as is possible, but it 
may not be generic enough...

The header is used to sign request parameters, including the parameters 
of the OAuth header.

The handshake is a multi-step process to get a user's credentials: 
request a temporary token, send the user to the site, user comes 
back and enters a code, request a permanent token.

That's all there is to it : )

Yeah ... I looked at this before and decided it was a little tricky 
:)

A more generic implementation may be a protocol that wraps around 
http(s).

read/custom oauth://[chris-:-api-:-twitter-:-com]/ [post a: 1 b: 2 c: 3]

sounds like a good idea :)

Doable, I think.  Need to have a way to store user info over sessions.

system/user ?

I think that's fine if it's for your own purposes...

as opposed to ?

Well, even for twitter, I maintain an account for my soccer team 
- so I have an app key/secret and two user key/secrets.

Then for google, you'd need an app key/secret and a user key/secret.

So you'd need to store by domain and user.

I see ...

api.twitter.com
 [
    "rgrebol" [
         key/secret/other metadata
    ]
]

docs.google.com
 [...]

And...

api.twitter.com
 [key/secret]
docs.google.com
 [key/secret]

get-env returns the value of an env variable, so you could concat 
that, system32, and the dll name

replace the lib loading line with this:


k32-lib: load/library join to-file get-env "systemroot" %"/system32/Kernel32.dll"

it works for me

does it work for you?

not quite - the correct path is still prefixed with the directory 
of the rebol exe

** Access Error: Cannot open /C/dev/r2/C:\Windows/system32/Kernel32.dll
** Near: k32-lib: load/library join to-file get-env

need a leading "/" to make it an absolute path, no?

oops... to-file should be  'TO-REBOL-FILE

OK, now

ok, will update rebol.org file.

I get:

>> do http://www.rebol.org/download-a-script.r?script-name=windows-chrono.r
connecting to: www.rebol.org
Script: "Chrono - High-precision time measurement" (24-Aug-2010)
** Script Error: Feature not available in this REBOL
** Near: i64-struct: make struct! [
    low [integer!]
    hi [integer!]
] [0 0]
>>

you are not using the latest rebol which has library support for 
free.

2.7.7 or higher?

2.7.7

core or view?   AFAIK only the view version has the library support.

Library support requires /View for now, not /Core.

that was core - with view, it works

what was the reason for limiting it to /View?

It was limited to /View before. That limitation was just not removed 
yet. 2.7.7 was put out quickly.

I get significantly varying results from consecutive executions of 
 print ".". Is this typical?

>> probe time-lapse [print "."]
.
0:00:00.000521096
== 0:00:00.000521096
>> probe time-lapse [print "."]
.
0:00:00.001658861
== 0:00:00.001658861
>> probe time-lapse [print "."]
.
0:00:00.001802788
== 0:00:00.001802788
>> probe time-lapse [print "."]
.
0:00:00.00178686
== 0:00:00.00178686
>> probe time-lapse [print "."]
.
0:00:00.001694701
== 0:00:00.001694701
>> probe time-lapse [print "."]
.
0:00:00.001815304
== 0:00:00.001815304
>> probe time-lapse [print "."]
.
0:00:00.001718594
== 0:00:00.001718594
>> probe time-lapse [print "."]
.
0:00:00.0017971
== 0:00:00.0017971
>> probe time-lapse [print "."]
.
0:00:00.001743625
== 0:00:00.001743625
>> probe time-lapse [print "."]
.
0:00:00.002421733
== 0:00:00.002421733
>> probe time-lapse [print "."]
.
0:00:00.002234001
== 0:00:00.002234001
>> probe time-lapse [print "."]
.
0:00:00.001800513
== 0:00:00.001800513
>> probe time-lapse [print "."]
.
0:00:00.001718594
== 0:00:00.001718594
>>

On Windows, yes. You will get varying results on anything. It's a 
non-real-time multitasking thing.

sure, I would expect some variation, but this seems a bit large

anyhow, this should be averaged over thousands of executions to get 
a more accurate value

Anything that requires traditional I/O (not OpenGL or DirectX) will 
have a lot of variation on Windows. Video games use timers and delays 
to get consistent behavior.

adrian, its up to you to put a loop in the block.

but, you can get a feel for relative speeds.... sine, on my system 
takes a little over 0.000015 seconds on average.  when I run the 
timer with a loop of 100000 It takes about 1.5 seconds... so the 
resolution of the timer is factual.

I'm using the timers to have much better resolution in controling/measuring 
frame rates.

its not perfect, because the time events aren't very reliable, but 
at least I can measure some of mouse events and throttle them.

does chrono-time work for you?