Subj : Re: kermit protocol in syncterm
To   : fusion
From : Digital Man
Date : Fri Aug 18 2023 19:02:13

  Re: Re: kermit protocol in syncterm
  By: fusion to Digital Man on Fri Aug 18 2023 08:23 pm

 > On 18 Aug 2023, Digital Man said the following...
 >
 >  DM> YMODEM-G is a protocl. gkermit is a terminal transfer protocol driver.
 >  DM> You're comparing apples and oranges. If you want to compare X/Y/ZMODEM
 >  DM> protocol performance, you should be comparing with the reference
 >  DM> X/Y/ZMODEM protocol implementation (rz/sz).
 >
 > sorry i misspoke. i can see gkermit (and by extension, kermit. just to spell
 > it out) outperforming sexyz's ymodem-g.

Okay. <shrug>

 > you asserted ymodem-g shouldn't have any overhead and should be as fast or
 > faster than kermit. is that in theory?

Correct, I'm talking about the protocol itself. Unless you're using a compression feature with Kermit (are you?), it's really impossible for the Kermit *protocol* to outperfrom Ymodem-G protocol over the same TCP/IP link - removing any implementation details. Ymodem-G simply has near zero over head over TCP/IP and no wait for acknowlegement of any subpacket data, so the only want to get *faster* is to use a protocol with literally zero data overhead (e.g. FTP) or to use some kind of data compression.

 > because i tested ymodem-g with sexyz
 > and it performs extremely poorly with incredible overhead. (again,
 > sexyz->syncterm)

"Overhead" refers to the amount of extra data that is sent in addition to the actual payload (file) data. It's not really relevant in that sentence. The overhead is a function of the protocol, not the impplementation. Anyway, okay, so you found gkermit outperms sexyz in your particular test environment. Cool.

 > if i compare to the reference implementation of rz/sz what am i trying to
 > prove?

Your initial messages said you were comparing Kermit to Zmodem. If that's what you really want to do, use the reference implementation of Zmodem by Chuck Forsberg for your tests.

 > that sexyz's ymodem-g (or zmodem) works better with the reference
 > implementation than with itself? or am i proving both perform poorly
 > compared to kermit?
 >
 > you mentioned zmodem shouldn't be far /behind/, yet it performs better than
 > ymodem-g with sexyz.

Faster than Ymodem-G? Not under my tests. And there's no technical reason why it should.

 > people are easily using it correctly and getting poor results where they
 > shouldn't be. so they try kermit and it blows sexyz out of the water. after
 > which you chime in and say "use ymodem-g it should be even better!" .. well
 > i'm saying that doesn't pan out in real life.

Your "real life" test is over a localhost link or a local network? Why on earth would you be using a serial/modem file transfer protocol over a local network (Ethernet, WiFi?) and call that "real life"?
-- 
                                            digital man (rob)

Sling Blade quote #2:
Karl (re: killing Doyle): I hit him two good whacks in the head with it.
Norco, CA WX: 83.6øF, 41.0% humidity, 11 mph ESE wind, 0.00 inches rain/24hrs
--- SBBSecho 3.20-Linux
 * Origin: Vertrauen - [vert/cvs/bbs].synchro.net (1:103/705)

.