April 2011 - Oberon - mailman3.inf.ethz.ch

Re: [Oberon] ETHO to *nix MTA interface?
by easlab-absa 02 May '11

02 May '11

> > it details mostly just the 'configuration' for > > someone who already knows the principles. > > OK, I'll be reviewing & improving the page. Well no, it's appropriately written for a higher level of knowledge than mine. > > AFAIK, on *nix machines any one IP-port can communicate > > with any other active one, if the configuration allows? > > What do you mean by IP-port? > > The Internet has sockets. > http://en.wikipedia.org/wiki/Internet_socket > In the most common terminology, > socket = ((IP address),(port number)). > Each protocol has a distinct port number as explained in > the Wikipedia. For example, a destination socket for SMTP > might be ((10.4.0.1),(25)). > Isn't: 10.4.0.1 = IP and 25 = port; hence IP-port? > A network adapter is configured in software as a network > interface. > > > So, for ETHO to communicate with the MTA, the ETHO > > application would need to communicate with the ISO-stack. ETHO talks to the ISO-stack at the application level, which passes the signal down to the socket-level, which talks to the MTA's socket, I'm guessing. > Are we discussing one machine or two machines? One. > Is ETHO running on a Linux machine which also runs an MTA? Yes > Are ETHO and the MTA on separate machines? No. In the second case, does "the MTA" belong to you? Is it on your premises? > > OTOH if I used a network-card, the appropriate ETHO > > facility [and presumably also for LEO] would communicate > > in tcp/ip ONLY at the output of the card. > > > > That's not what I need, for LEO to communicate with > > the *nix MTA. > > Why? Ethernet is fast and supported by *nix and by ETHO. Because there's no physical distance that the signal needs to travel [outside of a PC]. I suspect that LEO needs to communicate with: loopback/127.0.0.1, <smtp default port>/25, which is completely different from it communicating with V24/RS232. Thanks, == Chris Glur. I really must go from my land-line tomorrow, for a few days.

2 1

ETHO to *nix MTA interface?
by easlab-absa 29 Apr '11

29 Apr '11

Newsgroups: comp.os.linux.networking,comp.mail.sendmail Subject: How to interface a front-end to a MTA? I've got a nice front-end which I've been using for years, to access the Inet via a moden and my ISP. Because of email SendAuthorisation problems, at the ISP side, I want to 'route' via a linux MTA. P.E. wrote:- > The tunnel works well now and is described here. > http://carnot.yi.org/NetworksPage.html OK I fetched that with the 2 substantial daigrams, but it details mostly just the 'configuration' for someone who already knows the principles. At this stage I lack basic knowledge of networks. Eg. my idea of setting Oberon.Text to replace the <smtp.URL> which was designed for a RS232 -> modem output, is simplistically absurd. AFAIK, on *nix machines any one IP-port can communicate with any other active one, if the configuration allows? And the ISO-stack allows an application to 'communicate' down to the 'IP-port' level. So, for ETHO to communicate with the MTA, the ETHO application would need to communicate with the ISO-stack. But LEO currently only talks with the modem, and only at the ISP is tcp/ip interfaced? OTOH if I used a network-card, the appropriate ETHO facility [and presumably also for LEO] would communicate in tcp/ip ONLY at the output of the card. That's not what I need, for LEO to communicate with the *nix MTA. Apparently ETHO-VNC is related to this matter, because that also communicates INTERNALLY [on the one same machine] from ETHO to some port/s via tcp/ip? I want to understand the higher-level theory before I start 'just editing config files'. Thanks, == Chris Glur

2 1

ETHO and VPN tunnelling
by peasthope＠shaw.ca 28 Apr '11

28 Apr '11

Folk, This was a little over four years ago. >From peasthope at cablelan.net Thu Mar 22 20:30:23 2007 Date: Thu Mar 22 20:30:33 2007 > I'm still trying to make the SMTP/SSH tunnel work. >From wbh at conducive.org Thu Mar 22 21:37:53 2007 From: wbh at conducive.org (W B Hacker) Date: Thu Mar 22 21:38:13 2007 > No public-facing MTA that I am aware of will accept an 'always on' SSL tunnel on port 25. The preceding quotations are just a small sample of the discussion. The tunnel works well now and is described here. http://carnot.yi.org/NetworksPage.html = http://142.103.107.138/NetworksPage.html A few noteworthy points. * Cantor and Heaviside are old PCs running ETHNO. * The tunnel is between Dalton and Joule which run Debian Linux. * Cantor and Heaviside use the tunnel as if it were any ordinary network link. Neither needs TLS. * SMTP, POP3 and other protocols work through the tunnel as if it were any network link. The communications are not legible from outside the tunnel. * Cantor is about 50 km from the ISP WAN but communicates as if directly attached to the WAN. Thanks to everyone who responded to my questions. Certainly would have been more difficult without the help. Comments, suggestions and questions pertaining to the Web page are welcome. Regards, ... Peter E. -- Telephone 1 360 450 2132. bcc: peasthope at shaw.ca Shop pages http://carnot.yi.org/ accessible as long as the old drives survive. Personal pages http://members.shaw.ca/peasthope/ .

1 0

mailing list archive
by peasthope＠shaw.ca 28 Apr '11

28 Apr '11

Hello ETHZ, At the top of http://www-old.oberon.ethz.ch/native/archive/ is link "http://www.lists.inf.ethz.ch/mailman/listinfo/oberon". The link is faulty. Switching to https and removing the www gives a working link. https://lists.inf.ethz.ch/mailman/listinfo/oberon http://lists.inf.ethz.ch/mailman/listinfo/oberon also leads to the archive but the protocol should be evident. Regards, ... Peter E. -- Telephone 1 360 450 2132. bcc: peasthope at shaw.ca Shop pages http://carnot.yi.org/ accessible as long as the old drives survive. Personal pages http://members.shaw.ca/peasthope/ .

1 0

Complex ETHO mailArchive problem.
by easlab-absa 28 Apr '11

28 Apr '11

Using LEO to compose this 'log' while extracting info. from the relevant linux files, I see that: on Dec 1 2007 I used curl -o ObnMailLst \ https://lists.inf.ethz.ch/pipermail/oberon/2005/004110.html This would/should have fetched https://lists.inf.ethz.ch/pipermail/oberon/2005/004110.html to the file: ObnMailLst [https is a curse. I don't think ETHO handles it ?] Embedded in this file linux : mc <plain ascii editor> sees: AsciiCoder.DecodeFiles % js.TextPopups.Mod ~ 52J0_o1YJIa;M9jK7kLMYC;<A HREF="https://mailman.inf.ethz. Since the original name of the file is 004110.html I'll make a copy with that name and use a 'standard browser', opera to view it. I see:-- AW: [Oberon] USB Open Host Controller Driver, AosUsbOhci.Mod. sedlacek sedlacek at swissonline.ch Tue Jul 19 17:06:53 CEST 2005 ... Fax: +41 (81) 860 06 62 AsciiCoder.DecodeFiles % js.TextPopups.Mod ~ 52J0_o1YJIa;M9jK7kLMYC;kM9jK732 at 0;@F?Q>dgbfg:W3d1dX\=\dI0bdg>feFFb<07j ... NB in `opera`the string: kM9jK732 at 0 is a hotLink, and there are many more, all with the format <string> <space> "at" <space> <string> For some reason opera decodes the view of these hotLinks. Other hotlinks on the page/file look normal. OK, hotlinks need a view: field and a URL: field, and this corrupted file has got only spurious URL-fields. That's another reason why I HATE M$loth, for their wizard mentality and deceptively transforming the bytes, and breaking ASCII by introducing their own non-standard substitutes for the existing apostophe & back-quote. So that when M$ users write "can't", it looks like garbage. But now, for a 3rd opinion, via: EditTools.OpenUnix /home/crg/Inet/wget.curl/ObnMailLst ETHO, will show the actual bytes, which are:- Fax: +41 (81) 860 06 62 AsciiCoder.DecodeFiles % js.TextPopups.Mod ~ 52J0_o1YJIa;M9jK7kLMYC;<A HREF="https://mailman.inf.ethz.ch/mailman/listinfo/oberon">kM9jK732 at 0</A>;@F?`<A HREF="https://mailman.inf.ethz.ch/mailman/listinfo/oberon">kS at S</A>`Q>dgbfg:W3d1dX\=\dI0bdg>feFFb<07j ============= That's one line starting with "52J0" which the emailing may 'break'. At what stage between sedlacek's ETHO and mine was his ASCII corrupted [transformed for a better experience]? Thanks, == Chris Glur.

1 0

ETHO miscl
by Chris Glur 27 Apr '11

27 Apr '11

If I get the ETHO bounce-confirm from this [before I have to leave tomorrow], it means that my old mailAuthPatch still works, and my gmail account is barring me. Here's my trace [still left in the code as a mess, from ca. 2006] :-- send= 777 mailing250 HELP= 20 250 HELP= 31 TxAuthenticate 334 VXNlcm5hbWU6= 94 334 <passwrd>= 94 235 Authentication succeeded= 94 <-- *! Or possibly my ISP had changed the mailAuth protocol from the single one: LOGIN; which my patch handles. This chaos all happened around the time of the Chinese reportedly hit on google. And these days google is no longer hitting me with: <there seems to be a problem with your 'server' which is suspected of being infected with a denial of service virus, because your IP sends requests in rapid sucession, which a good normal user using WinTel IE, and being hypnotised by our dancing graphics couldn't do>. So that I had to intersperse my goog-fetches with other [time consuming] non-g-fetches in my <list-of-to-be-fetched-URLs-by-lynx> for text only fetches. And here's a quirky thing related to <being too fast> with a PC that I rescued from the garbage about a decade ago, when everybody was 'upgrading': -------- Newsgroups: comp.lang.pascal.misc, comp.lang.pascal.delphi.misc > I copied my TP7 directory to the Win7 thing, but I get: > error 200 divide by zero > when I compile/run the \examples\crtdemo.pas .. Sounds like you're using the Delay function which was software driven. Your PC maybe too fast. .. The offending code is used only to initialise the Delay function, which is part of the Crt unit. Since the Crt code is monolithic, putting the Crt unit in a uses statement causes the error. Of the machine is over about 200 MHz, the Delay function in the default Crt unit CANNOT be called. ----------end of extract. ------------------- Frans-Pieter wrote: > Linux Oberon I did not try-out yet. However as I will attend the talk of > Günter Feldmann http://www.oberonday2011.ethz.ch/talks/ it seems Previously Chris Burrows acknowledged the sublte and important fact that after you've worked on a system for some time, you subconsciously KNOW where the faults are, and you can no longer objectily test it. A problem with LEO, which I work around by not switching to another task, while LEO is fetching from Inet, might have a common-cause with the other <linux Oberon>: LNO; which I just recently rediscovered. In LNO I lose keyboard input ability when I switch out to another task [desktop, VT]. ----------- Of course it's more professional to use the existing linux MTA, than to keep patching the ETHO code for the next mailSendAuthorisation protocol. But configuring `sendmail` is evil punishment ! They've got a make file which manipulates the macro pre-processor, which manipulates the configuration file!! That's like tying your shoe laces, wearing boxing-gloves and viewing via a mirror. On the favourite topic of mine: 'splitting a task into independantly testable serially stages, I'd like to first confirm if I can get LEO to mail to my 'linux'. I'm guessing that an Oberon.Text with SMTP = "127.0.0.1" {* outgoing mail server *} might do something. ==> "no connection" Oh well perhaps Peter Easthope has research this. Or Frans-Pieter can enquire at the oberonday talk Thanks, == Chris Glur Oh don't forget to restore: Oberon.Text

1 0

Re. Oberon DAQ was: User report on LNO
by easlab-absa 25 Apr '11

25 Apr '11

Frans-Pieter wrote: > This line in your code made me curious: > RGtrace.LablVal > > I wonder if your oberon system works as a dataserver connected to a DAQ. > No knowledge of dataservers. CRGtrace.LablVal is just an external procedure to [IIRC] write a string and a number, like: "LoopCountNow= 7". > I'm looking for an oberon solution to gather data from a rs232 port, make > a simple presentation and send it trough ftp to a website.* > So you want to: [decomposing the total task] collect rs232 data to a file; manually manipulate it; send the edited data via ftp, where the destination irrelevantly needs to be a 'web site' ? This seems to be 3 independant steps. And I want to make 2 related comments [which don't immediately help solve your problem]: 1. this idea of serially-transforming-data, related to compositional languages, where each stage is independant is very economical of labour, and what I'm trying to morph ETHO towards. 2. The stage of "manually manipulate it" is the most important and expensive part, and that's where ETHO excels. > In the V24 code examples I saw a nice solution to put a task in the > oberonloop that reads the uart. > Why do you need to consider low-level aspects, like UARTs, if V24 can do serial transfer. > So my idea is to dust an old native oberon machine and install the > networking tools. > Did you trie out the ftp server lately? > No. > > The blackbox/windows, the linux-kst, the LNO or the Bluebottle >solutions seem to overcomplicate things. The other advantage of the 'compositional' approach is that, since the stages are mutually independent, you can do some in ETHO and others in linux ...etc., which is well demonstrated bellow by Peter's suggestion. Peter E. wrote:- > If a Mail Transfer Agent, MTA, is running on your Linux > system you don't move a text file from ETHO to Linux and > then send the emessage from a Mail User Agent in Linux. > Your ETHO system sends the emessage directly to the MTA > which forwards it to the ISP. It's just a matter of > configuring exim or Postfix properly on the Linux system. > I like that idea. Also, if I learned how to network-configure, I could do some 'local' testing of eg. send-confirmation, instead of dialing into my ISP to test each step. > Exim is still the stock MTA in Debian. I preferred > working with Postfix when it was necessary but it was an > optional installation. > This present installation has got 'sendmail' which is reputed to be a monster. I'll try to get some advice from the 'network USEnetGroup' how to configure it if I fail after reading the man. Perhaps I could extent that idea, to have an ETHO-PC connect to this Win7-HPnetbook, so that I could have inet wireless-connection, during the week, while I'm away from from a land line. The HP-toy has a [looks like 8 pin] socket that I think is called a "J<something>". If I got a network-card for the PC, could I perhaps interface with LEO and let the Win7 thing do the inet transactions? Then does the Win7, become a 'router'? I wonder if it can do that. Communicating between 2 PCs seems more naturally done by USB than a network, but I haven't yet found out how to do that. Thanks, == Chris Glur.

3 2

Re (2): User report on LNO
by easlab-absa 24 Apr '11

24 Apr '11

Because I need to use multiple unsyncronised systems to email, recently, I'm not sure if I've already sent this. From: easlab-absa <easlab(a)absamail.co.za> Date: Mon, 18 Apr 2011 21:37:47 +0200 > > I never succeeded with ETHO's Mail.Panel ... > > Why? What was the difficulty? That was when I started using ETHO: 1997?8, when P. Muller was very active. He tried to help me debug it. I was getting error-mesg from my ISPs server. Inet debugging is difficult because you have to depend on the line & the 'other side', and pay for the connection/time. Once I found that I could handle mail with 'commands' instead of Mail.Panel , I forgot about Mail.Panel . It won't work now because it uses `Mail.Send *` which hasn't got SendAuthorisation capability. > > ... without notification my ISP added mailSendAuthorisation ... > > Ideally Oberon users write support for it. As previously explained: I wrote code for the easiest of the 2 authentication-protocols which my ISP offered and tested/used it for a few years, until the ISP 'changed again'. > Otherwise set up any > old PC as a Linux based MTA which happily cooperates with ETHO. Yes, that's why I emphasise the value of ETHO which 'cooperates with linux'; but I HATE having to move this text to `linux's evolution' to mail it. Fortunately I can still receive with ETHO. == Chris Glur. PS. Because it's not solid-technology, it's difficult to explain: the POWER of ETHO as a research tool is demonstated by the following. How, *while still in the email text* I can cutNpaste [after easily finding it] the original code of my TxAuthenticate patch. ------------ PROCEDURE TxAuthenticate; PROCEDURE SendNexpect(Send: ARRAY OF CHAR; Expect: CHAR); BEGIN SendCmd(Mail,Send, ""); Response(Mail, s); CRGtrace.LablVal(s,94); (* Out.String(s); Out.Ln(); *) IF s[0] # Expect THEN Out.String("TxAuth error: "); Out.String(Send) END; END SendNexpect; BEGIN Out.String("TxAuthenticate"); Out.Ln(); SendNexpect("AUTH LOGIN","3"); SendNexpect("Y3JnbHVyQGFic2FtYWlsLmNvLnph","3"); SendNexpect("NT***Qz","2") END TxAuthenticate; ---------- Especially with the ETHO colouring which is lost in the emailing, I can immediately see that the protocol selected is "AUTH LOGIN", and I guess there's my UserName and password in Base64.

2 1

LNO & SendMailAuthentication
by easlab-absa 19 Apr '11

19 Apr '11

After I had a sleep it came back to me: TextMail.Mod is a seperate alternate to the more commonly used Mail.Mod . And because TextMail.Mod had a particularly nice, clean structure, I decided to extend TextMail.Mod to give me the required SendMailAuthentication facility. Per RFC [which I've got on an 'old' N-O partitions, which is why LNO's ability to read N-O partitions is so valuable] SendMailAuthentication works something like this: after contacting the smtpServer and getting a reply, the client sends "EHLO <hisEmailAdr>"; server replies with the types of SendMailAuthentication which it can handle; client selects one and uses it. Because there are about 6 SendMailAuthentication protocols all modern MailClients would need to handle ALL of them. The first time my ISP introduced SendMailAuthentication, Telnet showed that of the 2 protocols use, the easiest looking one used Bas64 encoding for the protocol. So for my limited use, I just did the encoding 'in the source code' for MY use, instead of having the object code do it every time. So I needed 2 different source & object file for my 2 mail-boxes. Now apparently my ISP doesn't handle the previous protocol. Probably M$ changed it so that their victims would all have to buy new stuff. Can you see why I won't d/l a big WinETHO file for my M$-only wireless-modem capable Win7 device? == Chris Glur.

1 0

Re: [Oberon] User report on LNO
by easlab-absa 18 Apr '11

18 Apr '11

On Sun, 17 Apr 2011, easlab-absa wrote: > > > > I'm writing this from LNO, which I got working again, after several > > years. > > [System.Time 15.04.2011 19:39:05 > > ETH Oberon / Linux Native 12.Jan.2002] > > [snip] > > > PS. writing this with ETHO was easy [hence the size]; > > but get it mailed via the abusive Wintel monopoly > > constraints is Not easy. Duke wrote:- > Here's what the header of your message says: > > X-Mailer: Ximian Evolution 1.4.5 (1.4.5-7) > I took a while to work out your point. Yes I WROTE 'that' mail in LNO. Like now I'm WRITING with LEO. Get it? I d/l-ed your mail via LEO and used <select> & `TextMail.Cite *` appropriately. > and here I thought that you had cranked out such this nice message using > Multimail or some other Oberon OS hack. Actually, I was hoping that > there was a long lost Oberon OS MUA that through your good graces , > was miraculously surfacing. :) If you think in terms of "long last" you set yourself up for disapointment like I suffered. We are NEVER finished paying the WinTel abusive monopoloists. There is no LAST payment with the Mafia. In 19xx when my ISP said Win3 would no longer suffice, and that I'd have to buy W95, I said screw-you and looked to linux & ETHO. I never succeeded with ETHO's Mail.Panel but I used `Mail.Send *' and <mailReceive> for several years. Then without notification my ISP added mailSendAuthorisation which was standard for M$ and had been added to most mail clients. So I had to hack <Mail.Mod> to make EASMailTxAuth.Directory EASMailTxAuth.Show EASMailTxAuth.Send * ..etc. NOW recently I discovered AGAIN that my emails were no getting, delivered; but the worst part of it was that, since I had 2 mail accounts at my ISP, I used to Cc the <other one> as a confirmation. Little did I relise that since the <other one> was in/on the <same server> it didn't NEED <to get out> and need sendAuthorisation to be delivered !! So now I send the Cc to my gmail address as confirmation. Which is also problematic. So now to mail this, I must, 1. paste this to /tmp/t17, 2. mark /tmp/t17, 3. EditTools.StoreUnix tmp17, 4. start <Ximian Evolution> garbage, 5. ...import..the text file /tmp/t17 6. bla..bla Life is not easy. You like most casual users, wrongly concentrate on what the conputer does, rather that what YOU [the most important part] does. It's the powerfull Human interface of ETHO that's important. Let me give an example: your post reminded me of Mail.Panel which I haven't seen for years; and I was wondering if it uses the same module/s which I had to hack to get CRGMailTxAuth.Send. So now RIGHT HERE in the middle of the email, I can examine what modules Mail.Panel uses: * klux Mail.Panel in a document = opens it, * mark the `Send *` [buttons], => Columbus.Inspect ~ == shows 'Mail.Send *' So that means the Mail.Panel also uses the same module which lacked SendAuthorisation capability. Obviously I'm very annoyed that the ETHO mailer again fails my ISP's new requirements, and I'm not keen to hack it again. Although I must admit, that I'm curious which of the several sendAuthorisation protocols it NOW uses. So occasionally I do a Telnet 'probe' to experiment. But all my 'research documentation' re. mail sendAuthorisation is on N-O partitions. Which is also why I was so enthusiastic to get LNO partly running; because LNO can read the N-O partitions while I'm working under LEO. LEO is much more convenient that LNO, having : EditTools.OpenAscii ^ EditTools.OpenUnix ^ EditTools.StoreAscii * EditTools.StoreUnix * IIRC you were using LNO & I asked about the latest version and got no reply. Perhaps that was during the period that I didn't realise that my mails were failing. Thanks, ==Chris Glur. I't good that you mentioned 'Multimail' PS. I'm editing this in <whatsItCalled> I just switched back to LEO, and confirmed that Multimail doesn't MultiMail.Send which unlike Mail.Send May have the sendAuthorisation facility. BTW, what was YOUR problem/complaint with emailing? BTW-BTW evidence that one is WRITING in <whatsItCalled>, and NOT ETHO, is that <whatsItCalled> has a spell checker. Sorry about the memory. I don't need it with ETHO, since everything is 'linked', PSS. what version of LNO do/did you use? Do you have the problems that I described? If not why didn't you provide feedback? PSSS I've just tried MultiMail: it give admin probibit TOO.

2 1