Mailing List Archive: 49091 messages
  • Home
  • Script library
  • AltME Archive
  • Mailing list
  • Articles Index
  • Site search
 

Proxy

 [1/8] from: luca:krenet:it at: 13-Jul-2001 12:27


Hi everybody, I have a trouble with a proxy. When I am at work I have to use a proxy and I can't configuring Rebol to read internet pages. I have tried all the possibilities ('socks, 'socks4, 'socks5, 'generic) but no one work well. The 'generic one ask me a username and a password and I don't know them. I've tried the LAN username and password but it doesn't work. At work I have a Windows98 system and this the reply of the proxy to a general request: HTTP/1.1 407 Proxy Access Denied Server: Microsoft-IIS/4.0 Date: Mon, 12 Feb 2001 07:31:25 GMT Proxy-Authenticate: NTLM I think the problem is the NTLM Proxy-Authenticate. The rebol version is REBOL/Core 2.5.0.3.1 Can someone help me to solve the problem? Thank you.

 [2/8] from: dockimbel:free at: 13-Jul-2001 13:01


Hi Luca, I have the same problem at work. Rebol can't do anything for you in this case because it does not support NTLM authentication (which is a Micro$oft proprietary scheme). Currently the only solution is to install on your pc the Microsoft Proxy Client which will tunnel your http connections thru M$'s Proxy Server. Here's some unofficial documentation on NTLM : http://www.innovation.ch/java/ntlm.html HTH, DocKimbel. Luca Truffarelli wrote:

 [3/8] from: luca:krenet:it at: 13-Jul-2001 14:34


At 13:01 13/07/2001 +0200, Nenad Rakocevic wrote:
>Hi Luca, > >I have the same problem at work. Rebol can't do anything for you in this case because it does not >support NTLM authentication (which is a Micro$oft proprietary scheme). Currently the only solution >is to install on your pc the Microsoft Proxy Client which will tunnel your http connections thru >M$'s Proxy Server. > >Here's some unofficial documentation on NTLM : http://www.innovation.ch/java/ntlm.html
Thank you Nenad. Do you think it is not possible to realize the "3 messages" scheme with rebol?

 [4/8] from: dockimbel::free::fr at: 13-Jul-2001 15:50


Luca Truffarelli wrote:
> At 13:01 13/07/2001 +0200, Nenad Rakocevic wrote: > >
<<quoted lines omitted: 8>>
> Thank you Nenad. > Do you think it is not possible to realize the "3 messages" scheme with rebol?
Yes it's possible, but there's a hard part : you have to implement DES and MD4 in Rebol which is not really a simple thing to do with Rebol scalar types. Their bit size is lesser than the C or Java equivalent, resulting in a painfull work to adapt the algos in Rebol. The 'Bignumber component should be the solution for such works as soon as RT will release some docs on it. The http scheme should also be modified to work with M$ Proxy Server, because MPS doesn't strictly obey to the http standard. (as always with M$!) Any volonteer to make a tunnel through the Death Star ? ;) Regards, DocKimbel.

 [5/8] from: luca:krenet:it at: 16-Jul-2001 15:09


At 15:50 13/07/2001 +0200, you wrote:
>Luca Truffarelli wrote: >>
<<quoted lines omitted: 18>>
>The http scheme should also be modified to work with M$ Proxy Server, because MPS doesn't strictly >obey to the http standard. (as always with M$!) Any volonteer to make a tunnel through the Death Star ? ;)
Now I can't, but maybe I will try later.

 [6/8] from: holger:rebol at: 16-Jul-2001 8:27


On Fri, Jul 13, 2001 at 03:50:55PM +0200, Nenad Rakocevic wrote:
> Yes it's possible, but there's a hard part : you have to implement DES and MD4 in Rebol
DES is already supported by REBOL/Command 2.0. We might add MD4 support in one of the next updates.
> which is > not really a simple thing to do with Rebol scalar types. Their bit size is lesser than the C or > Java equivalent,
No. REBOL uses 32 bit integers, just like most other languages. Integers are always signed, but for the types of calculations done in MD4 and DES this should not matter, except additions/subtractions have to be done carefully to avoid overflows. An alternative is to use a 16-bit implementation of MD4/DES. That gets you around the signed vs unsigned problem.
> resulting in a painfull work to adapt the algos in Rebol. The 'Bignumber component > should be the solution for such works as soon as RT will release some docs on it.
No, bignum uses variable-length integers, primarily for modulo arithmetics. It helps for RSA, DSA and Diffie Hellman, but is rather useless for DES, MD4 etc. Besides at the moment there is no bignum! datatype yet, so you cannot access the component from the REBOL layer.
> The http scheme should also be modified to work with M$ Proxy Server, because MPS doesn't strictly > obey to the http standard. (as always with M$!) Any volonteer to make a tunnel through the Death Star ? ;)
Do you have the specs for MS-Proxy in HTTP ? -- Holger Kruse [holger--rebol--com]

 [7/8] from: mario:cassani:icl at: 16-Jul-2001 17:24


Hallo Holger,
> DES is already supported by REBOL/Command 2.0. We might add > MD4 support in one of the next updates.
[...]
> > The http scheme should also be modified to work with M$ > Proxy Server, because MPS doesn't strictly > > obey to the http standard. (as always with M$!) Any > volonteer to make a tunnel through the Death Star ? ;) > > Do you have the specs for MS-Proxy in HTTP ?
Can you please give a /Core and/or /View solution and not only a /Command one? I don't mean to necessarily put MD*/DES to the free REBOL/* but to simply find a general solution: I think that a lot of people using the free versions is behind such an horrible kind of a M$ Proxy and this REBOL diffusion... Best regards Mario

 [8/8] from: holger:rebol at: 17-Jul-2001 9:54


On Mon, Jul 16, 2001 at 05:24:36PM +0100, Cassani Mario wrote:
> Can you please give a /Core and/or /View solution and > not only a /Command one? > I don't mean to necessarily put MD*/DES to the free REBOL/* > but to simply find a general solution:
A general solution to this would require MD4/DES in non-Command versions, significantly increasing the size of the binary, so, sorry, what you are asking for is impossible. Btw, it will take a while until we can support MSNT proxies in HTTP anyway, because their "protocol" requires HTTP keep-alive (and thus multi-level HTTP ports with connection caching as in FTP), which we don't support yet. It is planned in conjunction with the async HTTP rewrite though. At that time we may look at MSNT-HTTP as well. -- Holger Kruse [holger--rebol--com]

Notes
  • Quoted lines have been omitted from some messages.
    View the message alone to see the lines that have been omitted