[sebhc] list archive and a specific question

[Jim Battle]

Dwight K. Elvey wrote:

...
> Hi
>  I thought I'd mention, I don't do checksum. My experience with
> serial is that it is either working or it is not working.
> When I captured the images, I'd transfer them twice. A file
> compare shows any random error. I had no errors in any of the
> transfers except on the serial/cassette card H8-5. These
> failed at the 9600 baud to transfer complete images. They
> had a circuit that used optical isolators that just didn't
> make it for this high a baud rate. My program has the option
> to select a slower baud.

In the past, for my Sol and Wang projects, I did what you did -- send the file 
twice and diff it to ensure a good transfer.  It has always been a waste of time 
  to transfer it twice and to manually to and diff things, but I never felt 
comfortable enough just assuming the first pass is going to be good.  Even if 
the PC is fine, the old computer might have problems.  For instance, on the 
Sol-20 there is no provision for hardware flow control, so you just attempt 
transfers and if you get a failure, lower the speed and try again.  I'd rather 
have a 2x faster channel that has to resend 1% of the time than have to run at 
the lower speed.  Since my laziness algorithm is of a greedy nature, I never 
invested the time to write a more robust transfer protocol, and I kick myself 
every time I have to transfer programs again.


--
Delivered by the SEBHC Mailing List