World: r3wp

Yes, that gives quite extreme performance. It's an elegant way to 
solve the problem of caching that others are providing via expensive 
slap-on solutions rather than by design. This is also why I got the 
idea that you would be able to create VID-like applications in the 
browser, because data is always resident server side, like an app.

The session data problem gets reversed: You have to avoid data to 
spill over to the next user of the app. This could provide a very 
unique way to handle form data, by simply dumping POST data in a 
resident object. Then you can quietly decide what to do with that 
data without having to worry that it's forgotten at the next page 
load.


Session data is about storing that data in a context isolated for 
that user. I've not studied closely how Cheyenne handles session 
data, but I've been working a bit on the form issues for REBOL/Forum.

I can implement a clean-up routine for RSP variables declared in 
a webapp context, but this would be a partial solution (won't clean 
up global space), and in all cases, you *should* initialized all 
variables before using them either by declaring them in a function! 
as local words or by explicitely setting them to a default value. 
Such clean-up routine could be usefull to enhance security by avoiding 
to reveal other user data in case of a RSP script programming error.


Btw, you can already detect uninitialized variables in your RSP code 
by running Cheyenne with the -w 0 command line option. This would 
tell Cheyenne to use a single RSP worker process that will be restarted 
after each request (CGI like beahaviour). An uninitialized variable 
will likely error out in that case.

Henrik: session data context is here to ensure that your data will 
be :
- preserved from being lost whatever happens to RSP processes
- isolated from other users

Having a Cheyenne running locally using a browser window to display 
VID dialect looks like very doable. I think that even 'move events 
would work fast enough. That would solve a lot of current View/VID 
issues while providing a cross-platform GUI. Add a proxy service 
to Cheyenne, and you got a nice RIA platform with online/offline 
working capabilities.  Anyone rich enough here to sponsor such project? 
:-)

Dockimbel, that might solve some problems I had with form submission.


My intent with forms was to provide an easy way to have all form 
data provided by the server via an object. When you create a new 
object it would hold info for when the form was created and a unique 
ID for the form.

Through that you can tie a form instance to a specific browser instance, 
and when the form is submitted, you can do server-side verification. 
If the verification fails, the form object remains and the page is 
redisplayed. If the form object validates, then the form object is 
removed or copied away from the block of existing form instances 
and can no longer be used from that form instance, if you attempt 
to submit again. This would eliminate accidental double submission, 
although not regular spamming. By having that framework, setting 
up a flow for how to handle form data, server side, would be simpler.

This doesn't sound like so much, but I happen to have an HTML dialect 
around, where I can create forms as objects in a simple way, and 
applying actions or handlers to forms, makes it much more like programming 
a real GUI. It could probably scale down to single text fields and 
a bit of AJAX.

I would initially not try to do VID graphics directly, but to do 
something like this:

html-view [
	text (reform "The current server time is" now/precise)
	display-name: text
	name: field

 button "Submit" [set-html-face display-name get-html-face name] ; 
 'name is stored on server too
]

... well, let's see how easy that is in JS :-)

doc, for persistent variables I am storing them in the session object. 
 I just hadn't expected some values to persist outside the session 
object as well.

Now that I know, I can be more meticulous in parsing etc.

Anyway, I stumbled across the correct solution it seems :)

There should be big sign somewhere, that people should enclose the 
variables set by parse in context or as function's locals. And it's 
not only Cheyenne related.

I normally do this, but for some reaason I was thinking cgi instead 
of rsp, and so thought I didn't need to do that.

Right, and the same good practice should be applied to all others 
dialects that include REBOL code, like VID.

I updated the Syllable build recipes for Cheyenne and UniServe to 
the new versions

Nice, is it online somewhere?

So far, for the new versions, only the recipes for building from 
source:

http://syllable.cvs.sourceforge.net/viewvc/syllable/syllable/system/apps/utils/Builder/packages/uniserve-0.9.33/

http://syllable.cvs.sourceforge.net/viewvc/syllable/syllable/system/apps/utils/Builder/packages/cheyenne-0.9.19/

The previous versions are included in Syllable Server and the new 
Cheyenne will be included in the next Syllable Server:

http://web.syllable.org/Linux/downloads.html

Nice work.

This is a video of my RSP pages showing binary at the top of the 
page.  http://screencast.com/t/WD2j8PJtse

The binary appears on a new RSP session

It will go away if I restart Cheyenne

Is there anything I can do to debug what is going on?

Each time I go to the RSP page, more binary appears above my html.

I'm running an command  encapped version of 9.19 ( I presume there's 
no difference between the earlier 9.19 and the official release )

you can try it yourself ... http://www.compkarori.co.nz:8000

I suspect I did somethng while writing some buggy rsp code

- The garbage binary data looks like compressed data. 

- "more binary appears..." => issue with a string! or binary! buffer 
not cleared or initialized as literal value instead of using MAKE.

- "...above my html" => check your 'on-page-start event handler in 
%app.init.r

on-page-start: has [][
	set 't0 now/time/precise
]

REBOL [
	Purpose: "RSP environement init code"
]

on-application-start: does [
	;--- add here your library / modules loading
    *do %private/captcha.r
    captcha/set-fonts-path %private/fonts/
]

on-application-end: does [
	;--- add here your library / modules proper closing
]

on-session-start: does [
	;--- add here your per session init code
	;--- ex: session/add 'foo 0
	;--- that can be latter accessed with : session/content/foo
	
	session/add 'user "guest"
	session/add 'hits 1
]

on-session-end: does [
	;--- add here your per session closing/cleanup code

]

on-page-start: has [][
	set 't0 now/time/precise
]

on-page-end: has [pos time][
	if pos: find response/buffer "</body>" [
		time: to-integer 1000 * to-decimal (now/time/precise - t0)
		insert pos reform [
			"<br><br><small>Processed in :"
			either zero? time ["< 1"][time]
			"ms.</small>"
		]
	]
]

nothing special ...

This has happened a few times now .. I mentioned it before.  And 
usually only when I am rewriting RSP code.

Are you using encmd v2.7.6 for encapping?

yes

running on Linux?

windows

I need cmd for the odbc

If there's nothing I can do to debug this, I'll restart cheyenne

You can restart it, I've a copy of the output.

The garbage data seems to increase by 830 bytes each time.

I thought it might be an internal response/buffer corruption (not 
been clear after the request), but the compressed size of the page 
is only 442 (in deflate format).

You should carefully examine each use of response/buffer in your 
RSP code. See if you're not inserting in response/buffer a growing 
 series?

server restarted

the only response/buffer I have is ...

nothing

You should at least have one occurence for the captcha image

Need to go now, will give it a look tonight.

no, I do have this too ...
		response/buffer: xfdf

        response/set-header 'Content-type "application/vnd.adobe.xfdf"

ah :-)