On Wed, 2009-09-02 at 12:34 -0500, folderol wrote: > Unless I'm mistaken, z-modem is quite heavily error-checked, so losing > some characters could probably be absorbed without you noticing. > Yes, the Z-modem protocol uses block-level checksums and will re-request failing blocks. However I think I'd notice if it was re-requesting transfers: the program displays a running total of blocks transferred. This clicks up steadily and at about the same speed as it did on a Win95. I haven't noticed any pauses while its doing this, so I don't think I'm getting retries. Typical file sizes are 100-200KB and the transfers are done at 9600 baud. As the blocks transferred count is roughly the same as the file size in KB I assume its using a 1KB blocksize. During a transfer the count clicks over at around 4-5 blocks per second, which matches the baud rate and guessed block size, so its certainly not re-requesting every block - that would drop the rate to nearer 2 blocks/sec, which would look obviously slower than I see when a transfer is running. > The Application I'm using only checks the validity of entire strings, > and doesn't retry - it just aborts. > > Yeah, I know. Crap :( > I can't argue with that. As my logger downloads are running reliably at 9600 baud on a fairly slow computer (an 866 MHz P3 box) I don't think your problem is in WINE. Can you try using something else to do transfers over the link (e.g. minicom or Kermit) to check it isn't a hardware problem? Kermit would be ideal, provided you can install it at both ends, since its available for both Linux and DOS/Windows and does display error stats. Also, you could try both the Linux port and the DOS/Windows port under WINE at that end of the link. In addition the file transfer test can be scripted and tried easily at a variety of baud rates. HTH Martin
