Poke a Shell Variable with xTalk?

Sivakatirswami katir at hindu.org
Wed Sep 14 21:37:28 EDT 2005


Jim, interesting, though I'm not much closer to a solution with this,  
it did prompt me to "think different" find another way to state the  
question.  In OSX we have "open process" but we can't do that for  
Unix process on the same box except through shell..

Now on our Linux web server I use this CGI for sending mail from  
remote rev Apps, where those apps have no mail services, but simple  
POST a msg to the web server... Now, if I could just "translate" this  
to work from inside a stack on OSX, we would have it.  Or I might  
just use this CGI as is and put this CGI on an internal web site on  
our inhouse OSX Serve... but such a hack!   I would rather be able to  
do this directly from any box...

########

#!/usr/local/bin/revolution

## This little CGI simply avoids the need to install an SMTP library
## on any client rev app. Instead, I just have the client app build a  
post
## string, parse that and let my server's send mail do the rest.
## the TO is hard coded, but obviously doesn't need to be.
## this saves a lot of headaches debugging email services in a rev  
client app.
## and you can do other stuff with this if you want, log it, poke a  
dbase etc.

on startup
if $REQUEST_METHOD is "POST" then
   repeat until length(tDataIn) >= $CONTENT_LENGTH
     read from stdin until empty
     put it after tDataIn
   end repeat
split tDataIn by "&" and "="

## the following is what I've been trying to emulate from the  
"console" version of Rev
##  without success...

put "/usr/sbin/sendmail -t" into mprocess
         open process mprocess for write
         write "From:" && (urlDecode (tDataIn["from"]))& cr to  
process mprocess
         write "To:" &&   "katir at hindu.org"  &  cr to process mprocess
         write "Subject:" &&   (urlDecode (tDataIn["subject"]))   &   
cr & cr to process mprocess
        write    (urlDecode (tDataIn["body"])) &  cr to process mprocess
        close process mprocess
        put the result into tResponse

## in case this is being done from a web form then send back the  
response to port 80

     put "Content-Type: text/html" & cr
     put "Content-Length:" && the length of tResponse & cr & cr
     put tResponse

end if
end startup





On Sep 13, 2005, at 4:10 PM, Jim Ault wrote:

> This may help, then again, maybe it does not apply to your situation
>
> http://developer.apple.com/technotes/tn2002/tn2065.html
>
> Q: How long can my command be, really?
>
> A: Calling do shell script creates a new sh process, and is therefore
> subject to the system’s normal limits on passing data to new  
> processes: the
> arguments (in this case, the text of your command plus about 40  
> bytes of
> overhead) and any environment variables may not be larger than  
> kern.argmax,
> which is currently 262,144 bytes. Because do shell script inherits its
> parent’s environment (see the next question), the exact amount of  
> space
> available for command text depends on the calling environment. In  
> practical
> terms, this comes out to somewhat more than 261,000 bytes, but unusual
> environment settings might reduce that substantially.
> Note: This limit used to be smaller; in Mac OS X 10.2 it was about  
> 65,000
> bytes. The shell command sysctl kern.argmax will give you the  
> current limit
> in bytes.
>
> On 9/13/05 6:36 PM, "Sivakatirswami" <katir at hindu.org> wrote:
>
>
>> OK the above works, but I want to try now piping tMsg straight into
>> Send mail *without* saving or reading a file from the hard drive
>> (why? some new security thing in OSX, Postfix preventing more than
>> 1024 chars input without introducing a CRLF... even right in the
>> middle of a word, I'm getting a space in the middle of a word in the
>> HTML version a complete bad line break in the middle of a word (every
>> 1024 chars) in the text alternative)
>>
>
> or maybe you need to interact with the process directly as in
>
> Q: I have started a background process; how do I get its process ID  
> so I can
> control it with other shell commands?
>
> A: You can use a feature of sh to do this: the special variable $!  
> is the ID
> of the most recent background command, so you can echo it as the last
> command in your shell script, like this:
> do shell script "my_command &> /dev/null & echo $!"
> -- result: 621
> set pid to the result
> do shell script "renice +20 -p " & pid
> -- change my_command's scheduling priority.
> do shell script "kill " & pid
> -- my_command is terminated.
> _______________________________________________
> use-revolution mailing list
> use-revolution at lists.runrev.com
> Please visit this url to subscribe, unsubscribe and manage your  
> subscription preferences:
> http://lists.runrev.com/mailman/listinfo/use-revolution
>




More information about the use-livecode mailing list