...I wanted to let people know that my resident wizard managed to accomplish the impossible, and now I AM able to do zmodem uploads
with my still totally DOS telnetting setup!!!
...peering at TCP packet traces and comparing them to the ZModem and
TelNet documentation finally led me to realize the problem is caused
by the TelNet IAC character (0xff). ...TCPPORT doesn't escape the
IAC character on sends. It doesn't handle IAC sequences on receive
very well either. The ZModem part was working to specification...
It works a lot faster than the kermit I had been using... He's
uploaded his program to Chowdanet...
a DOS environment!... I tried telnet://guest.sailor.lib.md.us but there
is little hope since 'nix-like `ZMoDem' flavours are different about the Return Character via ~TelNet~ protocol, if i'm not mistaking... Well, i
Quoting MICHEL SAMSON from a message to NANCY BACKUS <--
Thanks a lot for making an *EFFORT* to share such good news,
Nancy!
And what a lot of sysops don't know that ZModem is one of the worst protocol to use over telnet because its timing is designed for POTS, not Internet. However, it was the most plentiful, so therefore, it proliferated even more.
And what a lot of sysops don't know that ZModem is one of the worst protoc to use over telnet because its timing is designed for POTS, not Internet.
i believe digital man has written SEXYZ to work with synchronet for windows. besides.. why aren't you guys just using NTP to begin with? :) it's with the times. truly external, too.
Also-are you still wanting me to try to get Kermit running
under my system again? Just wasn't sure if you did or not.
Quoting haliphax from a message to Sean Dennis <--
windows. besides.. why aren't you guys just using NTP to begin with?
isn't ymodem so much better?
isn't ymodem so much better?
Not really, it's designed for phone lines too and suffers the same timing problems as Zmodem.
windows. besides.. why aren't you guys just using
NTP to begin with?
What's NTP?
isn't ymodem so much better?
Not really, it's designed for phone lines too and
suffers the same timing problems as Zmodem.
windows. besides.. why aren't you guys just using NTP to begin
with?
What's NTP?
isn't ymodem so much better?
Not really, it's designed for phone lines too and suffers the same
timing problems as Zmodem.
Not really, it's designed for phone lines too and suffers the same timing problems as Zmodem.
All transfer protocols is designed for a phone line, but it bet it wouldn't make a difference over ip, why should it?
windows. besides.. why aren't you guys just using NTP to begin with?What's NTP?
Network Time Protocol, used for synchronizing clocks. In some environment where the clock is essential it's used, like in IP telephony.windows. besides.. why aren't you guys just using NTP to beginWhat's NTP?
with?
http://east1999.icednet.org/ntp is the url, if i remember. so far, you can only use it for downloads... but the next version is well underway. it's more or less a client/server based transfer protocol, independent of the bbs
Quoting Stephen Hurd from a message to Sean Dennis <--
YModem-g is actually well suited for for telnet, however it's missing
some nice feautues that ZModem has (ie: resume). No other YModem
should be used on high-latency links.
Quoting mark lewis from a message to Sean Dennis <--
zmodem is a streaming protocol unless you explicitly switch it to ACK blocks... zmodem's normal mode of operation is to not say a thing on receive unless there is a problem with a bad or missing block or
(yes) and overly long time period in which nothing is received...
Quoting Bo Simonsen from a message to Sean Dennis <--
All transfer protocols is designed for a phone line, but it bet it wouldn't make a difference over ip, why should it?
Ah, okay. I never was quite sure what YModem/G was used for. I haven't see it in many terminal programs that I've come across, except for Hyperterminal Zap-O-Comm and QModem Pro.
zmodem is a streaming protocol unless you explicitly switch it to ACK blocks... zmodem's normal mode of operation is to not say a thing on receive unless there is a problem with a bad or missing block or
(yes) and overly long time period in which nothing is received...
If that's the case, then why doesn't someone use the non-ACK mode for telnet connections? Every terminal program I've used seems to use that, even those that are designed for telnet (including ZOC, QModem Pro, HT)... doesn't make sense.
All transfer protocols is designed for a phone line, but it bet it wouldn't make a difference over ip, why should it?
As Mark explained to me, ZModem is using ACKs and that slows it down terribl
I was experimenting with Kermit when I was running Maximus/2 and it seemed t have the uncanny ability to run wide open with unlimited packet size.
As Mark explained to me, ZModem is using ACKs and that slows it down terr
I was experimenting with Kermit when I was running Maximus/2 and it seeme have the uncanny ability to run wide open with unlimited packet size.
ZModem alwaysuses delayed ACKs... the only time that Kermit is faster than ZModem is when it can compress the file using run length compression... so Z files do not transfer any faster for example.
Kermitas of last time I looked at it used a fixed packet size, ZModem uses a adaptive packet size... ZModem is also about 100 times easier to set up and use.
YModem-g is actually well suited for for telnet, however it's missing
some nice feautues that ZModem has (ie: resume). No other YModem
should be used on high-latency links.
Ah, okay. I never was quite sure what YModem/G was used for.
I haven't seen it in many terminal programs that I've come
across, except for Hyperterminal, Zap-O-Comm and QModem Pro.
zmodem is a streaming protocol unless you explicitly switch it to ACK
blocks... zmodem's normal mode of operation is to not say a thing on
receive unless there is a problem with a bad or missing block or
(yes) and overly long time period in which nothing is received...
If that's the case, then why doesn't someone use the non-ACK
mode for telnet connections? Every terminal program I've used
seems to use that, even those that are designed for telnet
(including ZOC, QModem Pro, HT)... doesn't make sense.
Kermitas of last time I looked at it used a fixed packet size, ZModem use adaptive packet size... ZModem is also about 100 times easier to set up a use.
Have you setup gkermit on your board? It's pretty simple. http://synchro.net/docs/kermit.txt :-)
Quoting haliphax from a message to Sean Dennis <--
you can only use it for downloads... but the next version is well underway. it's more or lessa client/server based transfer protocol, independent of the bbs.
Quoting mark lewis from a message to MICHEL SAMSON <--
kermit other than by you... are you still wanting to do/try something
with it on my OS/2 system, too??
Quoting Stephen Hurd from a message to Sean Dennis <--
need one that currently WORKS... either one that currently uses COM
ports or FOSSIL, or onethat seems to only work with some telnet
packages.
Quoting Stephen Hurd from a message to Sean Dennis <--
uses and adaptive packet size... ZModem is also about 100 times
easier to set up and use.
Quoting mark lewis from a message to Sean Dennis <--
Y-Modem-g and the other -g protocol(s) rely on the modem's
error-detection and correction... these protocols are faster since
they don't have to pollute the datastream with error stuffs trying to retrieve missing or corrupt blocks as the modem should already be
doing such...
Quoting Rob Swindell from a message to Stephen Hurd <--
Have you setup gkermit on your board? It's pretty simple. http://synchro.net/docs/kermit.txt :-)
Have you setup gkermit on your board? It's pretty simple. http://synchro.net/docs/kermit.txt :-)
Something you wrote?
Y-Modem-g and the other -g protocol(s) rely on the modem's
error-detection and correction... these protocols are faster since
they don't have to pollute the datastream with error stuffs trying to
retrieve missing or corrupt blocks as the modem should already be
doing such...
So I should set up a ymodem/g protocol in my board? :) I
wonder if CEXYZ/2 does it, actually... let me check...
need one that currently WORKS... either one that currently uses COM ports or FOSSIL, or onethat seems to only work with some telnet packages.
I've got two or three Pascal implementations... want them? :)
So I should set up a ymodem/g protocol in my board? :) I wonder if CEXYZ/2 does it, actually... let me check...
Nothing... and from playing with it a bit, it has a number of biggish issues...
If there's some way of getting touch with the authour, I'd be more than ha to help him fix stuff. But I haven't even been able to verify his existan as
of yet. :-(
as for the issues... i hate to burst your bubble, but it is running fairly well on at least 10 systems that i can think of personally. most of the issues you have are probably related to tying it in with the board itself a bit more, and he is addressing those with ntp 2.0.
I tried telnet://guest.sailor.lib.md.us but there is little hope
since 'nix-like `ZMoDem' flavours are different about the Return
Character via ~TelNet~ protocol, if i'm not mistaking...
The ZModem protocol doesn't use a return character.
If you can't interoperate with an Omen ZModem program, you're doing something wrong... and yes, it's probobly the IAC character.
...ZModem is one of the worst protocols to use over telnet because
its timing is designed for POTS, not Internet. However, it was the
most plentiful, so therefore, it proliferated even more.
I don't know about "straight" DOS, but I'd imagine that there might
be speed issues or protocol issues...
I only wish I knew more about protocols... I am pretty proficient
at Pascal, but I'm not as good as some of the people I know.
Also-are you still wanting me to try to get Kermit running under
my system again? Just wasn't sure if you did or not.
...are you still wanting me to try to get Kermit running...
...for that matter, i (finally) saw that you had uploaded a private
package to me, a long while back, of kermit stuffs... Are you still wanting to do/try something with it on my OS/2 system, too??
It would have been great to read Richard when this was On-Topic and
we had Ruurd Beerstra hanging around...
I think I was asking in there (with help from him) shortly after
Rurrd was there...
Have you setup gkermit on your board? It's pretty simple. http://synchro.net/docs/kermit.txt
Not really, it's designed for phone lines too and suffers the
same timing problems as Zmodem.
All transfer protocols is designed for a phone line, but it bet it
wouldn't make a difference over ip, why should it?
Because XModem and YModem both send a block, wait for confirmation,
rinse and repeat... this means that (for XModem) every 128 bytes
requires a round trip to be completed before the next 128 bytes are sent...
About "Pure DOS TelNet ZMoDem" of October 15:
Have you setup gkermit on your board? It's pretty simple. http://synchro.net/docs/kermit.txt
You shown how "charitable" you can be at times, lately, so i wonder
if you will agree to remove all mentions relative to me on `SynchroNet's ~WEB~ site - or at least, make them clear enough not to let people think
i actually approved `Kermit.INI'?
Should you feel like you're in one of
your good days, euh... it makes more sense to just point at the UpDate:
http://public.sogetel.net/bicephale/MSK.INI
The "Set File Collision OverWrite" line is one mistake i avoided in order to retain `Kermit's Transfer Recovery ("ReSend") feature, this one
is incompatible with Wayne Warthen's `Kermit for Windows' unless special care is taken but `ZMoDem' users wouldn't settle for less functionality.
"Set Retry 10", as far as i'm concerned, sure does sound like users
are given some time to launch their protocol but the delay which will be achieved via this option is going to vary with the `Kermit' packet-size.
A few more features have been overlooked in `Kermit.INI. Character Translation which makes it possible to use ~ASCII~ UpLoad for pre-writen text from a foreign environment isn't available. The Transaction Log as
it stands requires touch-up before anything is done, etc. Well, anyway!
windows. besides.. why aren't you guys just using NTP to
begin with?
What's NTP?Network Time Protocol, used for synchronizing clocks. In some
environment where the clock is essential it's used, like in IP
telephony.
haha.. funny that it parallels what we're talking about, but no. i was referring to the nemesis transfer protocol. i've been buggin the
author for a while to change the name because of this definition
conflict. :)
Quoting Bo Simonsen from a message to Sean Dennis <--
All transfer protocols is designed for a phone line, but it bet it
wouldn't make a difference over ip, why should it?
As Mark explained to me, ZModem is using ACKs and that slows it down terribly.
I was experimenting with Kermit when I was running Maximus/2 and it
seemed to have the uncanny ability to run wide open with unlimited
packet size. Unfortunately, I have not been able to figure out how to
get Kermit working under Telegard/2.
All transfer protocols is designed for a phone line,
but it bet it wouldn't make a difference over ip, why
should it?
Out-of-order packet reception springs to mind.
In the good old days of TeleNet (not ~TelNet~), the users were able
to adjust `ZMoDem' (in a similar way as with `Kermit') but most terminal emulators no longer give access to such long forgotten options and hence
the difficulties encountered by DialUp SoftWare using a ~TelNet~ "shim".
Because XModem and YModem both send a block, wait for confirmation, rinse and repeat... this means that (for XModem) every 128 bytes requires a round trip to be completed before the next 128 bytes are sent...
Yes but it's doesn't make them more optimized for modems than for telnet.. : Or does it?
Quoting MICHEL SAMSON from a message to SEAN DENNIS <--
`MS-Kermit.EXE' works with `Telegard/2' under `OS/2', instead of `CKOKer32.EXE'? It was tempting to replace `C-Kermit v5A(191)' of
(since i didn't expect ~TelNet~ using `SIO'/`VMoDem' to redirect data
to ~COM1~). A "Named Pipe" or else was mentioned but i heard you had
Quoting Bo Simonsen from a message to Sean Dennis <--
Doesn't it support external protocols like max?
Send 'em to deuce@bbsdev.net
For <deuce@bbsdev.net>, Site (bbsdev.net) said: 550 5.1.1 <deuce@bbsdev.net> User unknown
Quoting Stephen Hurd from a message to Sean Dennis <--
Argh... stupid... mutter...
Because XModem and YModem both send a block, wait for
confirmation, rinse and repeat... this means that (for XModem)
every 128 bytes requires a round trip to be completed before
the next 128 bytes are sent...
Yes but it's doesn't make them more optimized for modems than for
telnet.. : Or does it?
My round trip time to vert.synchro.net is currently 645.339 ms.
That is the time between EVERY 128 bytes in an XModem transfer... so
say I want to transfer a 5k file... 5*1024 = 5120 5120/128 = 40 40 * .645339 = 25.8
This means that 25.8 seconds will pass while my vert is just sitting
there waiting for my system to acknoledge the previous 128 bytes... 5k
can be transferred by both systems in under 1 second.
Doesn't it support external protocols like max?
Yes, but not at all like Maximus. I am using CEXYZ/2 with it, which
runs extremely fast and smooth. I'd set up Kermit just as an
experiment. :)
Sysop: | digital man |
---|---|
Location: | Riverside County, California |
Users: | 1,034 |
Nodes: | 15 (1 / 14) |
Uptime: | 21:17:45 |
Calls: | 778 |
Calls today: | 3 |
Files: | 95,174 |
D/L today: |
57 files (27,309K bytes) |
Messages: | 299,479 |