AltME groups: search

but we have yet Vincent Ecuyer that  make an MD2 convertion script 
that retrieves the pages from de dokuwiki on line and write on the 
local hard drive the file containing the infos converted to MD2 format

Nice tutorial Brian.

One way to notice very fast if you are reusing  words is to start 
a session with
  protect-system
or add it to your user.r file.
It can prevent stoopid coding errors --  like the time I wrote
    return: true
rather than
    return true

(that failed much later than you'd expect, and in a very strange 
way)

battery-file: to-file join battery-dir [ battery "state" ]

seems incorrect

Graham - What is a better way to create the battery-file filename?

well, if you did this

battery-file: join battery-dir join battery "state"

you wouldn't need the 'to-file

Actually, I don't think I need 'to-file in the first place.

Okay, i went ahead and changed "to-file join ..." to "rejoin [ ..."


Also threw in a little check for zero bit in calculating remaining 
time.

Do this in a REBOL prompt to run the test:

do http://home.tiscali.dk/john.niclasen/postscript/postscript.r

write %test.ps postscript load http://home.tiscali.dk/john.niclasen/postscript/test.txt

Now the file %test.ps holds the postscript output.

Do this in a REBOL prompt to run the test:

do http://home.tiscali.dk/john.niclasen/postscript/postscript.r

write %test.ps postscript load http://home.tiscali.dk/john.niclasen/postscript/test.txt

Now the file %test.ps holds the postscript output.

Do this in a REBOL prompt to run the test:

do http://home.tiscali.dk/john.niclasen/postscript/postscript.r

write %test.ps postscript load http://home.tiscali.dk/john.niclasen/postscript/test.txt

Now the file %test.ps holds the postscript output.

Pekr: from the geomol's link: "A PDF file is actually a PostScript 
file which has already been interpreted by a RIP and made into clearly 
defined objects. These objects are viewable on screen not in code, 
but in visual objects that everyone can see."

Pekr, what about making a small REBOL program, which can load postscript 
dialect txt-file, convert it to a .ps file, then call ps2pdf and 
finally call a PDF-viewer (Acrobat Reader) to see the result. You 
could also include a button to send the .ps file to a printer. Then 
you'll have PostScript with preview and all (also under Windows).

(Also I'm on OS X, so I don't have the preview problem, because the 
OS auto convert to PDF, if I'll like to see a .ps file.)

So, generate the output as postscript, and then use gsprint to print 
the postscript file to the epson printer ( default windows printer 
).

because eps is just a special postscript file that does not permanently 
alter the environment

I think I need to specify this in the PS file, otherwise the printer 
just prints everything in Letter/portrait

you need a bounding box in your ps file

There is a bug with printing multiple pages on certain HP printers. 
After the first page, some char is missing, that causes it to wait 
until you press the Online button or page feed button. I really would 
like that to be fixed. The work-around is to print to a file and 
have it printed using Printfile under Windows.

Added PageSize and images to the PostScript dialect. There seem to 
be a problem with jpg!?
Henrik, try this:

do http://home.tiscali.dk/john.niclasen/postscript/postscript.r

write %imagetest.ps postscript [page [translate 100 600 scale 72 
72 image logo.gif translate 100 500 scale 144 144 image logo.gif 
translate 100 300 scale 288 288 image logo.gif]]

Then view or print the file imagetest.ps

Yes, images are handled as REBOL image! types. If the dialect is 
given a file! or url!, it'll load the image. It can also take a word!, 
which is the image. It then convert the image/rgb to base-16 and 
put it in the ps-file.

So images in postscript files take up a lot of space! Most people 
knows this already. I tried to put a 100k png image in a postscript 
file, and it then was > 11MB.

I have problem with an image like in this example:

do http://home.tiscali.dk/john.niclasen/postscript/postscript.r
img: to-image layout [box 20x20 red box blue]

write %imagetest.ps postscript [page [translate 100 400 scale 72 
72 image img]]


Try print (or view) the imagetest.ps file. Do you also get an error?

No.  Just a debugging function that you add to your postscript file 
that you are trying to print.

It seems like, I got image to work. Try this:

do http://home.tiscali.dk/john.niclasen/postscript/postscript.r

write %palms.ps postscript [DeviceRGB page [translate 100x400 scale 
72x72 image http://www.rebol.com/view/palms.jpg]]


The file palms.ps can now be viewed or printed. If anyone got problem 
with this, please let me know.

will test with a local file soon

perhaps 12 seconds with a local palms.jpg file

It seems to work without the newlines, but then the ps file become 
difficult to enter with e.g. vim. That part is fast, I think. It 
only put in a newline for every 80 chars.

it does not, I tried a local file. 12 seconds.

You say, you tried a local file, but did you change the load-image 
in postscript.r?

so I think there should be an option to leave that out, or by default 
not have it. I'm guessing most people want to generate and print 
PS file, rather than read and edit them by hand.

now GS won't read the PS file, but the generation is much much faster. 
less than 0.5 secs

so for the other bug: I mentioned earlier that some printers I tried 
it on directly, will cause the printer to pause after the first page 
is printed. Then I need to press paper feed or reset the printer 
to get the rest out. Using Printfile for Windows solves that, but 
that program is really old and I don't want to rely on it in the 
future if my project is moved to Linux clients. I suspect there is 
just one char missing at the end of the PS file, but I don't truly 
know what causes it.

what I do to cause the error is make a ps file with postscript.r 
and then feed it directly to the printer. In windows I print directly 
to LPT1: to produce it. Under Linux I print it via LPR.

Output from TextEdit printed to a PS file has this in the end:

%%Trailer
%%Pages: 1
%%EOF
^D


The last char (ctrl-D) is hex 04. I think, the comment lines are 
according to "PostScript Language Document Structuring Conventions 
Specification". You can find this document on the net at Adobe. So 
there is a difference in how %%Pages comments are handled. And postscript.r 
doesn't put hex 04 in the end. You could try some of these changes, 
when you have the problem.

there is another method which is to get the bbox information from 
the font metric file itself. it's much faster, since it's just lookup, 
but I couldn't find an example of how to read it.

so my current method is Graham's, except for calculating the entire 
alphabet. we should probably work on getting that information from 
the font metric file instead.

The bbox information for the entire font is stored in each font metric 
file as a [llx, lly, urx, ury] coordinate set, so I shouldn't need 
to calculate it. This information is crucial in order to get one 
line of vertically centered text. I already got everything in place 
except that particular number. :-)

I've also seen a Laserjet 4500 lock completely up, when you feed 
it a specific postscript or PDF file.

did you try only on localhost ? can you add something like "127.0.0.1 
domain.local" in your /etc/hosts file and in the http.cfg add a corresponding 
domain preference

only occuring when I include javascript with the include-file directive

I'm only doing 

<% include-file something.js %>

are you passing a file! type to INCLUDE-FILE ?

<html>
<head>
<% include-file %button-header.js %>

	<title>Welcome to TestApp web application</title>
</head>
<body>
<img src="logo.png">
<center>
<% include-file %button-form.js %>
<a href="logout.rsp">Logout</a>
</center>
</body>
</html>

I take it the module order in the config file is a type of position 
as well?   if the tirst module returns true, then the rest not considered?

exactly, modules are competing to close each phases, so order in 
the list, and order in config file matter.

this case is solve only by the position in config file.

rsp would include the file correctly, but the generated page would 
reference libraries outside the webroot

Tested here, seems ok, I got an RSP catched error telling me that 
the file is not found.

can we log this to a local log file ?

the 'no-window option will trigger file logging mode

it is tested in cheyenne.r startup code and if defined, it will try 
to catch all errors in a local %crash.log file

Nenad, proably not necessary to have a method to restart modules 
after modifying... simplly dumped the code into an external file, 
and 'do that instead.

Customizable HTTP error pages: server-side redirection on 4xx and 
5xx errors to any pages, user-defined in config file.

If you want to try this method for windows.. here's what you do.. 


1) download the windows binary from openssl.. ->   http://www.slproweb.com/products/Win32OpenSSL.html

2) Unzip.. and pull out the openssl.exe file from the bin folder.. 
. drop that file into your cheyenne www folder
3) Create a self-signed cert....
3a) run openssl.exe

3b) enter this line: req -x509 -nodes -days 365 -newkey rsa:1024 
-keyout localhost.pem -out localhost.pem
(localhost is the cert name)

3c) answer the questions... when asked 'who are you?' enter your 
domain,  or 'localhost' as I did 

This will generate the cert in your www folder (this is just a demo... 
the openssl server uses it's location as root www folder)

4) Start up the server... enter this line into openssl:  s_server 
-accept 443 -cert localhost.pem -WWW


Now open any file in your Cheyenne www folder using the https:// 
 protocol

you only need the openssl.exe file for this demo

Does the rsp evaluate files included using include-file ?

-T
    Transparent proxy mode.


    If your machine supports it, stunnel will operate in transparent 
    proxy mode. Thus it will connect to the destination machine as if 
    it were the remote/client. In other words, the service to which you're 
    connecting will see the actual source IP address, rather than the 
    machine upon which this stunnel daemon is running. Helpful to maintain 
    good logs.

-p file

How would I send a binary file to a client?  Do I set up the correct 
http content headers, read/binary on the file, and then print it?

Isn't this just a mime-type issue?  Get the extension/file-type and 
let the browser
handle the download off a link?  Or am I missing something?

when the client clicks on the download, I have to retrieve the file 
from outside the web space

im guessing...  

req/out/headers:  List of [name [word!] value [string!]] header pairs
req/out/content: to-binary read file

Graham: look in %docs/rsp-api.html, see in Response Object / buffer, 
you'll have an example of the correct way to send an image file to 
the browser (just set the correct mime type for your own file).

try it in REBOL console, download and save the tiff file and try 
to open it to verify that your image is ok.

It doesn't seem that FF can handle TIFF files natively, I don't have 
the QT plugin so FF just propose me to download the file. I can see 
the TIFF file using my favorite imaging app, so no problem with the 
TIFF image.

there was even an image viewer plugin for Windows which would inadvertently 
clear the data whenever you would read pics off you camera!! hehe 
talk about incompetency!  yep, the plugin would open the file read-write, 
seek up till the end of the image data and close the file... which 
would then truncate it  :-) hahahaha

Cheyenne release v0.9.13 beta. Download at http://softinnov.org/tmp/cheyenne-r0913.zip

Changelog :


 o Session cookie management refactored. Cookies are now cached in 
 memory. 
	  Fixes all issues related to timezone.
	  

 o HTTPd request pipeline refactored. It's now more reliable, little 
 faster and

   able to handle extreme situations (stressing with FasterFox). Several 
   core parts
	  of Cheyenne have been touched, so watch out for regressions.
	  

 o Added 'forward method to RSP's Response object (now possible thanks 
 to the 
	  new pipeline engine).
	  

 o Internal modules API changed : 'deferred? property removed (not 
 needed 
	  anymore with new pipeline).
	  

 o Added 'on-status-code option in config file (see example in %httpd.cfg).
	  

 o do-sql now catches internal errors, so they can be more easily 
 located in 
	  calling context.

Regarding TIFF images, I have no control there.  Incoming faxes are 
saved as TIFF images and stored in the file system.  TIFF because 
they are often multipage images.  I am just serving them out again.

Cheyenne release v0.9.14 beta. Download at http://softinnov.org/tmp/cheyenne-r0914.zip

Changelog :

o response/forward improved : 

 - fully supports URLs as argument (can now forward to another virtual 
 host).
	- URL validity check (must have an explicit target).
	- protection against cycles.


o Command line option -p extended, now you can specify several listen 
ports separated
   by a comma (ex: -p 80,10443).


o New command line option -e : load and initialize Cheyenne without 
entering the

   event loop (needed for embedding Cheyenne in third party apps).


o Added a new experimental module: mod-embed. Purpose is to allow 
easy Cheyenne

   integration in third-party REBOL applications that require an embedded 
   web
   server. (Uncomment mod-embed in httpd.cfg file to activate it)


o Added %embed-demo.r file to show a sample of the mod-embed usage 
and API. 

o RSP: <% without %> eats all the memory. Fixed.

o URL-encoded request values were not parsed correctly. Fixed.


o RSP: fixed a typo in 'decode-params blocking the multipart data 
decoding and
   also a local word ('type) leaking in GC.

o UniServe's service startup refactored to be more flexible.



The new mod-embed is experimental. Please look at the %embed-demo.r 
file and send your feedbacks here.

I think that embedded mode might be the killer. Especially if you 
can combine it with encapsulation on a virtual file system

Encapsulation with virtual file system : that's planned for next 
week ;-)

Maarten, put your order in the httpd.conf file as well.

There's also the WORD part in modules, allowing you to add config 
options to the httpd.cfg config file, but currently undocumented.

Doc.. I had older PHP so now I have php-cgi listening, but it returns: 
"No input file specified". I have full path in the root-dir, is there 
something else?

Btw, mod-action provides a general purpose background tasks launcher, 
you could use it for making your first "bg handler". Here's a small 
sample code for implementing a bg task handler called "demo" :

- Add in %httpd.cfg in globals section :

	bind-extern demo to [.dem]

- Add  %handlers/demo.r :

REBOL [
	Title: "DEMO handler"
]

install-module [
	name: 'demo
	
	on-task-received: func [data][	
		data: reduce load data
		wait 0:0:3				; simulates a 3sec long processing
	

  result: reform [		; you have to return the response string in 'result
			<html><body>
				"You're IP is :" data/ip
			</html></body>
		]
	]
]


- Then create a fake %www/test.dem file (with any content, not used 
in this demo handler)

- Launch Cheyenne and try : http://localhost/test.dem

always requires a login

 : no, just remove or comment the 'auth "..." part  in config file.

application = folder (application name) + definition in config file.

MikeL: good idea, btw, I plan to add a "Reload Config File" option 
soon in this menu.

Pekr: AddHandler can be implemented now in Cheyenne. Should it work 
like in Apache (binding a handler to a file type) or can we make 
something smarter. Any ideas about that ?

Doing a simple test doesn't seem to work, there's maybe a regression 
on file uploads handling, give me a few minutes to test that.

To test file uploading, here's a short HTML form (put it in %www/) 
:

<form action="show.rsp" method="POST" enctype="multipart/form-data">
	<input type="file" name="ufile" size="16" />
	<input type="submit" name="usubmit" value="upload" />
</form>

BTW, file uploads are currently stored in memory only. I'll improve 
that to use the disk instead for big files.

What happens when I upload a 300mb file to Cheyenne (hypothetically?)

Currently, you'd better have a lot of RAM ;-). File upload support 
is still quite primitive, I'll improved that for the v1.

I am thinking of an S3 scenario where I want to "pipe" a huge file 
upload to Amazon S3 (as example)

Blocking for the end user : sure as long as he hasn't upload all 
the file, he won't receive a response from Cheyenne (that's the HTTP 
protocol).

Cheyenne release v0.9.15 beta. Download at http://softinnov.org/tmp/cheyenne-r0915.zip

Changelog :

v0.9.15 - 20/06/2007


 o RConsole was not started by default in the previous release. Fixed
	

 o RSP: 'include function protection from infinite cycles changed. 
 It's

   now based on a counter (5 maximum recursive includes). It's a little 
   less

   cleaner than stack-based tracking but much more reliable (avoids 
   matching
	  paths and targets).
	  

 o HTML.r library rewritten from scratch. Now, faster and more conforming 
 to 

   standards (Full range of Latin1 entities supported). Fixes URL-encode 
   bugs.

	o BugFix for command line parsing in encapped Cheyenne on Linux.
	

 o Fixed an issue with 'decode-multipart in RSP.r. File upload should 
 work ok 
	  again.
	  

 o Added a new global function : 'rsp-log value. Outputs values in 
 console for 
	  debugging RSP scripts. Works as 'probe.
	  

 o Reloading config file now supported. Running sessions and client 
 connections

   survive to the reloading process (needs some additional testing). 
   Activating
	  config file reload is done using:
	  
	  		- (Windows) "Reload Config" menu option in systray icon.
	  		- (UNIX)     kill -s HUP pid
	  		

 o UNIX signals SIGINT,SIGQUIT,SIGTERM now catched to allow cleaner 
 exit and last

   minute actions. Triggers the new 'on-quit event for HTTPd modules.
	

 o HTTPd internal events (not phases) refactored to be cleaner. New 
 module's
	  events added:
	
			- 'on-started:	when Cheyenne starts.
			- 'on-reload:   before a config file reload happens.
			- 'on-reloaded: after a config file reload happens.
			- 'on-quit:     when Cheyenne is about to stop and quit.
			

 o RSP sessions can now be made persistent (can survive to a server 
 complete restart).

   This option is controlled by a new config keyword: 'persist. Usage 
   is :
	  	
	        persist [sessions]    ; other flags can be added at will
	        

 o BugFix in session cookie handling for web-apps using 'auth mode. 
 Now the cookie

   is sent on the 302 redirection to the login page avoiding the creation 
   of a 
	  "shadow session" that will never be used. 
	  

 o FastCGI is under heavy work so mod-fastcgi is commented in config 
 file to avoid

   fastcgi startup. If you want to play with PHP, just uncomment the 
   line.

I have this odd problem with trying to execute a batch script from 
an rsp page.

I do a call/wait and get a return value of 0, but the script doesn't 
actually do what it's supposed to do .. viz. convert a ps file to 
pdf.
but the same batch file called from a rebol console does work.

I put an echo now > when.txt in the .cmd file and it definitely is 
being executed

but the same .cmd file works fine in a rebol console.

otherwise the cmd file would not work :)

and since it executes the batch file .. it can't be that.

if the batch file calls something of higher privileges. I don't know 
how it works though.

I start cheyenne up as a batch file using firedaemon, which turns 
this batch file into a windows service.

If I don't use firedaemon, then the batch file using 'call works. 
 But if I do, it fails :(

Check your httpd.cfg file to see if you find a 'post-max directive.

Sorry, no diff file list. You can update just the files having a 
more recent timestamp.

locally define your DNS names in your hosts file (/etc/hosts or C:\WINDOWS\system32\drivers\etc\hosts) 
to be able to simulate locally virtual-hosts.

Pekr, you have an issue with your localhost DNS resolution. You should 
check that your hosts file is not messed up. If you can fix the localhost 
problem, here's a workaround, in Cheyenne archive, edit the mods/mod-fastcgi.r 
file, find the fastcgi://localhost:9999 expression and replace it 
by : fastcgi://127.0.0.1:9999, that should work.

content of windows/system32/drivers/etc/hosts file:

127.0.0.1       localhost
::1             localhost

- ping localhost works in shell window
- print read dns://localhost works in rebol
- change in mod-fastcgi did not help ...