Re: [LHA-misc] Line parsing routine

[Jay H. <jh...@fa...>]

At 09:10 AM 12/1/00 -0500, you wrote:
>Paul Fox wrote:
>> 
>>  >
>>  > Without the size of the file we then need to change the way we download
>>  > the file to the Ocelot (which might not be a bad thing). Here is an
>>  > example:
>>  >
>>  > set addr=Ocelot:0.0.0 value=5 value=^@^M^J^Z^D
>> 
>> i feel like i'm starting to contribute to something i don't know enough
>> about, since i confess i haven't been following development closely.  but
>> if the protocol has been chosen as an ascii protocol for ease of debugging
>> (which is a big advantage of the smtp/ftp style of handshake) then i really
>> think you shouldn't be sending binary data as binary.  there are lots of
>> good encoding schemes to choose from.  remember -- you're going to want
>> to test this from telnet session.  and what you'll have in front of
>> you is a paper document from a device mfgr that gives hex codes.  very
>> convenient to be able to type 'value=\00\0d\0a\5a\04".
>
>You seem to be keeping up with everything pretty well. It was a nice point
>to note that if we needed compression on the fly that the file size
>would hinder things. Because of my networking background I tend to think
>that compression is best left to storage. If we compress data over a
>dialup we lose bandwidth because most modems use compression to get the
>extra bandwidth. From the tests I've done, I've seen the compressed file
>grow to 1.5 times the compressed size.
>
>I'll be honest, my thinking slipped. I want the hex-ASCII (hey what is
>correct name for that) for easy debugging. It works well and isn't much of
>a resource hog, although it does eat a little bit of bandwidth. But like
>Microsoft, I think bandwidth is infinite :-) (Hey! Put down those rocks
>I was joking!).

Forget the rock! I've got a stack of 40meg HD's that we use as door stops ;)

Actually, if you want to do downloads the best standard method around
that we use every day to download code to DSP's, embedded processors,
etc is Motorola S-Record format. It has the advantages that
- It is standardized in format
- Text/HEX printable based for dealing with 7-bit transmission
- Internal addresses/continuation/etc
- Internal checksums
- Standard line lengths (not to exceed 80 chars incl cr/lf)
- It can be decoded/cached in the receiving process and when the
   S7/S9 record format arrives it can be validated and d'lded as
   one piece.
- Most compilers can generate it.

My 2 cents.

Jay