World: r3wp

Wow thanks guys - I look some of these up

OCaml has some interesting features along those lines.

Doesn't smalltalk kind of work like that, since there are no keywords 
and everything can be redefined on the fly?

Except that being a pure object environment, you have a lot of small 
methods that accept very specific messages, so you don't really create 
dialects, though you can take the first step and create a vocabulary. 
It just doesn't get used the same way. I'm not a real Smalltalker, 
only tinkered a bit a long time ago, so someone correct me if I'm 
wrong.

That makes sense, Gregg. The difference between vocabulary and dialect 
can be hard to describe. Can Rebol dialects actually change the way 
the Rebol language functions?

Yes. Any Rebol code can change the way the language functions:
>> copy: func [a] ["hijacked"]
>> b: copy "123456"
>> b
== "hijacked"


As dialects are implemented in Rebol they can change the way the 
langauge functions.

For that matter, much of how REBOL functions now is actually implemented 
as REBOL dialects, some in native code, some in REBOL code.

See, for example http://www.rebol.net/projects/view1.3/chat65.html
(around about msg 25)

dialects are interpreters, which you can write in rebol. as interpreters, 
they can change everything. from rebol they inherit datatypes/syntax, 
which saves a lot work and makes things consistent. and the binding, 
which makes it easy to have variables in dialects and share them 
with rebol. the nice thing it, they keep their context. if you just 
pass strings and use regexp, you can do dialects in perl too, but 
the strings don't keep their context. which means sharing namespaces/locals 
with interpreters is less comfortable upto impossible.

I am working on a "REBOL powerpack". A set of libraries to help developers 
jump-start development. Check the REBOL powerpack group for details.

Funny question: What's a good iconic metaphor for "Save file"? Okay, 
we're used to have a floppy-disk-icon to click on, but do all people 
now-a-days remember what floppy disks were? I can't imagine a working 
successor, though. An USB-stick pixeled on a 16x16 pixel grid?  Not 
too easily recognised, I assume ... Any suggestions?

Funny answer:  should one really have to specify that something is 
to be saved?  rather than try to guide everyone from the desktop 
to the pictures of things that are really actions (floppy isn't floppy 
it's SAVE).  Why not assume that something is needing to be saved, 
and only bother the user about such details for exceptions, like 
to save under a different name or a different place than whereever 
it was you were keeping it.  What is the object that means the verb 
SAVE AS?  icons are generally pretty miserable unless they are representing 
Nouns.  use the word if you can.  otherwise stick with what is usually 
used, the floppy.

Good argument. Icon metaphors often are pretty misleading or ambigous 
at least. But the intention of my question was more of the kind of 
"What icon will we see in the future in places where now-a-days we're 
still used to find a floppy disk?" I'm sure we'll have to wait a 
reasonable amount of time until my computer knows what I want to 
save and what not ;-) We'll use icons for some more years, I suppose.

yeah, I think we are sorta stuck with it.  From a usability standpoint, 
deviating tends to be a bad thing.  most people will look for floppy, 
or know what it means when they see it.  since floppy's aren't good 
for much else, enshrining them as a token of how far we've come is 
fine.  there aren't good icons for actions.  we need arrows and things 
to show transference for most verbs, and 16x16 just doesn't cut it 
for showing some visual diagram of the action.  a document and a 
folder and an arrow showing movement of document to folder has been 
done quite a few times, and that one is probably closer than floppy, 
but when icons try to represent actions like this, the only semantic 
truly expressed is that there is movement of some thing.  and with 
files and folders or something, all the icons look the same except 
for the direction of the arrows.

thumb, as in hitchike, nmemonic for thumbdrive and transport ion

brain

optimist

Sadly we have not developed a lexicon of common icons.


The icon for power on/off is becoming standard (a circle with a vertical 
line through 12 O'clock)


The save icon being a floppy also looks like a SD card or Microdrive 
to a lot of people.

It would be nice if the standards were set up for everyone:

Save
Load
New
Quit
Prefs
Edit

Brain, Graham, would probably be better suited for wild angry clicking 
when recognising "I only thought I've saved!" :-D

Why do we have to save ?  "redo" action, you know ?  In real life, 
you don't save, you take one thing, you do something, and you stop. 
 When it can be the same in an app, it's better I think.  I have 
an app, where there is no save-button, this is very cool !

I agree Charles -- I also wrote an application in REBOL that has 
no Save button....the data set is consistent every time a field is 
exited. It makes for far fewer mistakes by the users.


But, when they do make mistakes, the applicaton has an Undo button. 
And that works even if they've exited and restarted the application.

That's right. Persistent data. It's the future (even though the idea 
has been around for ages.)

Hey Sunanda, how did you implement that... Carl's Rebodex has a nice 
little engine for doing the auto-saving, but I've been thinking of 
providing an undo feature as well.  Would be interested in seeing 
how you did it if your app is available for public viewing.

Brock, the app has an API to its own "flat file" database ... mainly 
blocks of REBOL objects.

On each update transaction (which usually affects several flat files) 
we record the before images.

That makes Undo really just the same as a Rollback.


The Undo log keeps the last 100 (or a user selectable number of) 
transactions, so they can go back a long way if they need to.


But,. like a lot of things, the code was experiemental and as such 
is a pile of kark. One day, I might tidy it up into an API undertstandable 
by someone other than me,  It the meantime, I that's enough high-level 
clues to help you replicate the logic.

hi rebols, is there any flag to set the number of max. redirections 
for the http-scheme ?

Sunanda, thanks, that does give me a good idea as to what you did... 
thanks.

Brock, also trap the window close event and any other way of ending 
your app, and, at that point, save to the default image. A "Save 
As.." button can be used to save to a particular file if the user 
wants.

Thanks Anton, that's how the Rebodex app works so I was aware of 
this method.

ok cool..

could anyone suggest me please free sntp (time protocol) server, 
we could test our device against? Thanks ...

ntp.xs4all.nl

ntp.surfnet.net

Mozilla Firefox will use bsdiff and bspatch for software updates. 
http://www.daemonology.net/bsdiff/, it remains me of xor based patching 
Gabriele came up with in the past. I wonder if similar aproach is 
still possible for Rebol products?

rebol is small enough, that you don't usually need it. :)

bzzz ....

:-)

Once again we will be wasting, even if we don't necessarily need 
to? :-) But we want fmod in rebol, tasking, font-engine, VM, extensions 
... it will not be small-enough for ages :-)

MAybe could be used to run scripts which insists on a special version. 
then the current one could be patched with a few-kb-download :)

petr: how much do you save if you use bsdiff from 1.2.1 to 1.2.223?

the waste is the memory and time used by bsdiff and bspatch in this 
case.

all you need, is using a smart autoupdate system like the one in 
the detective :P

I don't know how much do I save - you tell me. But Rebol slowly grows 
- View now has 700 KB, on GPRs connection, it takes few minutes :-)

patch size: 294kb, compressed rebol.exe 431kb.

is that worth the effort of telling the user how to use bsdiff? or, 
even worse, implementing that in rebol...

why is patch size so big? You once showed some 20kb or so patch size 
for an app. It probably depends upon what you actually patch, no? 
If it is rebol itself, it will nearly mean downloading it once again, 
it if it change in your encapped app, it could be smaller, no?

Normally, VM should be lighter when the dual rendering core will 
be removed: actually, there is the old Rebol one (for effects and 
text) and the AGG one (for Draw). Effects needs to be proted on AGG 
AFAIK

proted=ported

yes, that is possilbe, DideC

petr: don't be fooled. my technique was to update apps produced with 
THE SAME ENCAPPER, i.e. same version of rebol, changed rebol code.

so, you just update the rebol code.