>> For example, if you do it manually versus pasting it, does it affect
>> the syndrome?
> No. Well, sort of. After it's booted, at the login prompt, on the
> machine that usually (not always) doubles the letters, I see this on
> pasting:
> I paste: jklowden
> I type a space
> I paste: hello
> Total output: jklowdenjklowden àhelloxello
> Several pastings of "hello":
> helloxellohelloxellohelloxello
> Typing hello at different speeds:
> hxeelllloohxeelllloohxeelllloohxeelllloo
If this is manual typing, I doubt any human can type fast enough for
the time between characters to be anything but effectively infinite.
It looks to me as though h gets misechoed as x and space as à.
Looking at them in binary, assuming à is 8859-1,
01101000 h
01111000 x
00100000 space
11100000 à
The other characters involved are d e j k l n o w:
01100100 d
01100101 e
01101010 j
01101011 k
01101100 l
01101110 n
01101111 o
01110111 w
Adding start and stop bits and switching from MSB-to-LSB order to
temporal order,
0001001101 d
0101001101 e
0000101101 h -> echo as 0000111101
0010101101 j
0110101101 k
0001101101 l
0011101101 n
0111101101 o
0111011101 w
0000001001 space -> echo as 0000001111
This begins to look as though it might have something to do with
characters beginning with too many 0 bits. If it's something you can
try easily, I'd be curious to see what you get echoed if you type - at
human speed, ie, slow enough that each echo is long finished by the
time the next charcter arrives - the ASCII set in order:
!"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~
In particular, I'm wondering about the binary patterns of the corrupted
characters. The above makes it look as though 4 or more 0s are
problematic, leading me to wonder about space and !08@ABCHPX`ahpx in
particular (some of them have 0000 within them, others have 000 in
their low three bits, leading to 0000 when combined with the start
bit).
If it's easy to test, I'd also be curious what happens if you halve or
double the baudrate and then try the same test. In particular, if a
slower baudrate makes it significantly worse, I strongly suspect some
kind of power issue, because that makes it sound as though the line
being in spacing state for too long is problematic, where "too long" is
a fixed amount of time not related to the size of a bit time.
/~\ The ASCII Mouse
\ / Ribbon Campaign
X Against HTML [email protected]
/ \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B
_______________________________________________
Soekris-tech mailing list
[email protected]
http://lists.soekris.com/mailman/listinfo/soekris-tech
