[Tcl-bugs] [ tcl-Patches-420060 ] Add -address option to http::geturl

[<no...@so...>]

Patches item #420060, was opened at 2001-04-29 21:13
You can respond by visiting: 
http://sourceforge.net/tracker/?func=detail&atid=310894&aid=420060&group_id=10894

Category: 28. http Package
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Bud Beacham (bbeacham)
>Assigned to: Jeffrey Hobbs (hobbs)
Summary: Add -address option to http::geturl

Initial Comment:

Issue:
 
Linux allows multiple IP addresses to be bound to a 
single NIC card.  This gives someone the ability to 
write applications that appear to generate traffic 
from multiple clients from a single PC.
 
Problem:
 
The TCL library function HTTP::GETURL only uses the 
first IP bound to the computer.
 
Solution:
 
HTTP::GETURL uses the TCL socket command to setup the 
connection.  Socket has an option, -myaddr, that 
allows an IP address to be passed to the socket.  This 
will be the IP addresses used for the conection.
 
I modified HTTP::GETURL to allow an IP address to be 
passed to it.  It then gets passed to socket.  Now I 
can generate HTTPS GETs from a number of unique 
clients while using a single PC.

Note:  After I posted this on USENET someone showed 
how to use http::register to set the system IP to a 
differant IP and then that would be the IP used by 
http::geturl.  However, that is not as clean and 
simple as having an address option.  Also it creates a 
lot of CPU overhead having to call it each time you 
loop through a range of IP addresses.

Modified TCL 8.3 ...../lib/tcl8.3/http2.3/http.tcl 

proc http::geturl { url args } {
...deleted stuff....
# Process command options.

array set state {
-blocksize      8192
-queryblocksize 8192
-validate       0
-headers        {}
-timeout        0
-type           application/x-www-form-urlencoded
-queryprogress  {}
-address        {}       ;# added to allow addresses
state           header
meta            {}
currentsize     0
totalsize       0
querylength     0
queryoffset     0
type            text/html
body            {}
status          ""
http            ""
}
set options {-blocksize -channel -command -handler -
headers \
             -progress -query -queryblocksize -
querychannel -queryprogress\
             -validate -timeout -type -address}
...deleted stuff......

# this is where the enhancement patch executes
# I am unsure if the first "if" needs to be modified 
or not

 if {[info exists phost] && [string length $phost]} {
 set srvurl $url
 set conStat [catch {eval $defcmd $async {$phost 
$pport}} s]
 } elseif {$state(-address) == ""} {  ;# changed else 
to elseif - system default address, option not used
       set conStat [catch {eval $defcmd $async {$host 
$port}} s]
   } else {     ;# added else, set statement, and -
myaddr $state(-address)
         set conStat [catch {eval $defcmd $async {-
myaddr $state(-address) $host $port}} s]
     }

Example:

set result [catch {http::geturl $url -timeout $TIMEOUT 
\ 
 -headers "Cookie $cookie($localIp,$url)" \
 -address $localIp} token]
   
Documentation:

::http::geturl url ?options? 
	
The ::http::geturl command is the main procedure in 
the package. The -query option causes a POST operation 
and the -validate option causes a HEAD operation; 
otherwise, a GET operation is performed. 
The ::http::geturl command returns a token value that 
can be used to get information about the transaction. 
See the STATE ARRAY and ERRORS section for details. 
The ::http::geturl command blocks until the operation 
completes, unless the -command option specifies a 
callback that is invoked when the HTTP transaction 
completes. ::http::geturl takes several options: 

-address addr (borrowed from the -myaddr description 
in the socket documentation and added this part: "or 
multiple addresses bound to a network interface")
	
Addr gives the domain-style name or numerical IP 
address of the client-side network interface to use 
for the connection. This option may be useful if the 
client machine has multiple network interfaces, or 
multiple addresses bound to a network interface. If 
the option is omitted then the client-side interface 
will be chosen by the system software. 




----------------------------------------------------------------------

>Comment By: Andreas Kupries (andreas_kupries)
Date: 2001-08-24 10:56

Message:
Logged In: YES 
user_id=75003

Seems not, given the quietness.

----------------------------------------------------------------------

Comment By: Jeffrey Hobbs (hobbs)
Date: 2001-05-21 16:32

Message:
Logged In: YES 
user_id=72656

This is a good idea, but can we get it reduced to a simple 
patch against the current sources for easier analysis and 
inclusion?

----------------------------------------------------------------------

You can respond by visiting: 
http://sourceforge.net/tracker/?func=detail&atid=310894&aid=420060&group_id=10894