World: r3wp

A refinement is latched onto a function, so that you know that the 
refinement is part of that function. Hence, you must type out the 
function name, followed directly by the refinement without spacing. 
Then you type the function arguments and after that, the refinement 
arguments.

The arguments has the same order with refinements after the non-optional 
arguments:

with a function that requires 2 arguments and have 2 refinements 
should be used as:


myfunc/ref1/ref2 arg1-to-func arg2-to-func arg-to-ref1 arg-to-ref2

@Henrik  Thanks for the examples ..

@Endo     That's what I needed - a definitive HOWTO.  IMHO, the Help 
system should be worded that way as well, in order to sync with actual 
usage

If you study some unloaded data:

a: [now/precise now /precise]


a/1 is of type path!, which can be reduced to calling NOW with the 
/PRECISE refinement.


a/2 is of type word!, which can be reduced to calling NOW without 
any refinements.


a/3 is of type refinement!, which doesn't reduce. In principle, that 
refinement could be an argument to the function. What types you can 
pass to a function is almost entirely free-form.

>> reduce a

== [25-Oct-2011/17:39:39.218+2:00 25-Oct-2011/17:39:39+2:00 /precise]

The problem with wording the help differently is that arguments and 
their orders passed, might be different, depending on which refinements 
you use. Therefore help is formatted the same as function headers. 
A way to understand it, is that anything that comes directly after 
the function name is obligatory. When the first refinement appears 
in the help string, anything after that is optional.

Example:

my-func/boo 6

Does the 6 argument belong to the function or the refinement?

I'm getting TOTALLY confused here! I'm going to have to stop here 
and go study this - including all of you guys' advise - a bit more 
....

ok :-)

Yeah, that can be confusing! :)

>> f: func [arg /ref1 ref-arg] [print [arg /ref1 ref-arg]]     
>> f/ref1 /ref2 /ref3
ref2 ref1 ref3

>> f/ref1 'arg 'ref-arg
arg ref1 ref-arg

Duke: Don't be confused.

my-func/boo 6


If my-func requires an argument and /boo refinement does not require 
and additional argument, then 6 is for my-func.

if my-func and /boo both require args then you will get an error. 
my-func expects more arg.

if my-func doesn't require an arg but /boo does then it belongs to 
/boo

I personally consider refinements to be one of the less stellar parts 
of REBOL.


Henrik, do you prefer one additional function for each new way of 
calling a function, instead of refinements?

I should have made my example more like this:

>> f: func [arg /ref1 ref-arg] [print [arg ref1 ref-arg]] 
>> f/ref1 /ref2 /ref3
ref2 true ref3

I faced a problem when I use a refinements, my function had an ref. 
/any

Then I forgot it and use ANY native. It gives error about using none, 
because the function wasn't called /any, so the word ANY was defined 
and its value was NONE. It was difficult to find the problem.

Maybe refinements for functions are more clear with examples like 
this: Let's say, we want a sine function, which default operate with 
radians, but you can give degrees refinement, if you like, exactly 
opposite of the normal SINE function:


>> sin: func [value /deg] [either deg [sine value] [sine/radians 
value]]
>> sin pi / 6
== 0.5
>> sin/deg 30
== 0.5

Another example with additional argument, when refinement is used. 
The normal COPY function:

>> copy "abc"
== "abc"
>> copy/part "abc" 2
== "ab"

And checking help for COPY:

>> ? copy
USAGE:
    COPY value /part range /deep

Read that as:
- COPY takes one argument called VALUE

- If you choose to use the /part refinement, you then need to give 
an additional RANGE argument

- If you choose to use the /deep refinement, no additional argument 
is needed.

@Geomol  I now understand the syntax of function arguments, function 
refinements and THEIR arguments. Endo summarized it well. However, 
I still think that the REBOL Help sub-system should reflect EXACTLY 
how the function should be written. Thta should be the programmer's 
first line of help. Like the Unix man pages. So, assuming that the 
* char indicates optional syntax, IMHO, ?copy should say:
USAGE: 
    COPY */part, /deep* VALUE */deep range* 


To me, that tells me exactly the order in which the optional refinements, 
the required function argument, and the optional refinement argument 
should be written. To me this makes more sense. :D

COPY */part, /deep* VALUE */deep range*

 - not sure I understand this order, as it will fail to explain where 
 VALUE belongs and where RANGE belongs (it does not belong with /DEEP). 
 Anyhow, the notation is standard and should not change anywhere. 
 You only need to learn it once, so hopefully, this is not too much 
 of a hurdle.

I still think that the REBOL Help sub-system should reflect EXACTLY 
how the function should be written

 - this is easy. If you find a better way how to display the help, 
 then simply propose it. In REBOL, it is easier to write the code, 
 than to invent what the code should do.

But, do not forget, that it must be something that describes what 
do you want, not a description what you do not like.

It would certainly suffice if you wrote a couple of examples, how 
it should look

Aha, you tried to write something above. Well, I do not quite understand 
the stars you used - where should they be put?

The REBOL help system is remarkable concise and acts, usually, as 
an excellent aide memoire.


It has weaknesses too -- for example, it does not give hints as to 
which refinements are incompatible with others. eg HELP TRIM gives 
no clue that TRIM/HEAD/WITH is not acceptable.

A fresh look at the HELP system may help improve it.

Nor I did understand how can I recognize that VALUE is a parameter 
of COPY, not of the /DEEP refinement

The

    COPY value /part range /deep

actually means, that you can use any of:

    COPY value
    COPY/part value range
    COPY/deep value
    COPY/part/deep value range
    COPY/deep/part value range

(I must admit, that I never tried some of the above)

eg HELP TRIM gives no clue that TRIM/HEAD/WITH is not acceptable

 - yes, but that can be cured even without a change to the help system, 
 it might suffice to just add that as a new information into the main 
 help string.

@james_nak: thanks for your Oct-15th 8:33 PM reply.

Is it possible to open a web page and fill in username, password 
and press send? It is a POST form. I'm thinking of using it to check 
if some (of my own) web pages work and it is possible to login. I've 
tried read URL.

Yes: http://www.rebol.com/docs/core23/rebolcore-13.html#section-8.6

And more here: http://www.rebol.com/docs/core23/rebolcore-13.html#section-13.5

Interesting. Thanks, Geomol.

About the example 8.6 on read/custom and post. I can't find any info 
about the arguments for post. There is no help for this word. Can 
I send several arguments at once or do I write one post line for 
each form field?

You concatenate the args.
Example: as a GET

   r-get: read http://www.rebol.org/st-topic-details.r?tag=domain//html&start-at=26&limit=25

Same request as a POST:

r-post: read/custom http://www.rebol.org/st-topic-details.r[post 
"tag=domain//html&start-at=26&limit=25"]

Thanks, Sunanda.

Hello Helpful person.  I have a need for a local no-server database 
function and have explored the sqlite tools which I can get working 
OK But.  The data I want to be storing needs to be secure.  I have 
looked at the encodings.r and that will work but I was wondering 
if there was another way.  I have used Tsunami Records Manager in 
the past with good success.  I have used Euphoria which has an excellent 
inbuilt database function and I would like something similar if possible. 
 Any clues?

When I examine the functions in altjson.r and altwebform.r written 
by Christopher Ross-Gill, I see function definitions like below:


load-json: use [...<lots of words>...] [... <lots of definitions 
etc.> func [...] ]


is this method for hiding details in function body? or to make the 
function body cleaner?

Basically, it's so that the words in the USE are not easily accessible 
from the outside. A "poor man's module" if you wish.

I see, I thought that its a good way to keep the body clean. But 
curious about if there is another reason. Thank you.

Btw, I loved functions in utility.r. Is it 1.22.1 (4-Feb-2005) the 
latest version? fortype, nforeach, import & export are very useful 
functions.

yes, that's the latest version.

This is a good place to ask questions about getting started with 
REBOL and getting used to AltME.

Is there a LDAP module with authentification available for R2? I 
tried ldap-protocol.r from softinnov.org, but didn't get that to 
work (with anonymous bind) ... Not sure why I get "Error: Invalid 
port spec". I'm on Win7 (x64). Maybe it is something on OS level 
I have to config to use ldap://!?

Invalid port spec

 means that you have provided an incorrect argument to OPEN native.

Thnx, can I not use IP-address for argument?

Sure you can, but your syntax is maybe wrong, can you post here your 
connection opening code?

Hmm, I think I found what was wrong ... *blushing*

Anonymous login with ldap-protocol.r should work fine.

Now I got connected at least.  I used include instead of do.

Is there a way to loop throu the objects in LDAP? I seem to be able 
to get just one object at the time.