openssh unix dev - Jul 2015 - [PATCH v2 1/1] paint visual host key with unicode box-drawing characters

A machine diff between one using ASCII and UTF-8 would yield not equal?

Is it reasonable to only use UTF-8 ("pretty" chars) with a command
line
switch (forced) rather than as an override (implied) behaviour?

lists at wrant.com on Tue, 2015/07/07 05:43:
> A machine diff between one using ASCII and UTF-8 would yield not equal?
True. Do you need diffs of the artwork?
I suppose this broke several times in the past...
> Is it reasonable to only use UTF-8 ("pretty" chars) with a
command line
> switch (forced) rather than as an override (implied) behaviour?
Just another idea... We could print unicode characters only when a real
terminal is detected.

On Mon, Jul 6, 2015 at 10:43 PM,  <lists at wrant.com>
wrote:
> A machine diff between one using ASCII and UTF-8 would yield not equal?
>
> Is it reasonable to only use UTF-8 ("pretty" chars) with a
command line
> switch (forced) rather than as an override (implied) behaviour?
Please, lordie, yes. Unicode is *not your friend* for any kind of
binary or, in this case, display comparisons. It multiplies the
complexity of the relevant code.

Jim Knoble <jmknoble+kleankanteen at pobox.com> on Tue, 2015/07/07
05:58:
> It's not difficult to set LANG=C (or LANG=POSIX) in ssh's
enviroment. That
> should force ASCII.
LC_CTYPE=C should be sufficient.
> Otherwise, UTF8 should be used wherever possible (and
> LANG permits), not just for a terminal, IMHO. 
What comes to mind is pstree. Running it in a terminal you get a tree of
unicode characters, piping (or redirecting) you get ascii characters.

Basically it is a matter of preference. I would prefer ascii for a
non-terminal to make sure you can view a redirected text file without issue
even if unicode is not supported.
-- 
main(a){char*c=/*    Schoene Gruesse                        
*/"B?IJj;MEH"
"CX:;",b;for(a/*    Chris           get my mail address:   
*/=0;b=c[a++];)
putchar(b-1/(/*               gcc -o sig sig.c && ./sig   
*/b/42*2-3)*42);}
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 473 bytes
Desc: OpenPGP digital signature
URL:
<http://lists.mindrot.org/pipermail/openssh-unix-dev/attachments/20150707/c328edd2/attachment-0001.bin>

On Mon 2015-07-06 22:43:18 -0400, lists at wrant.com
wrote:
> A machine diff between one using ASCII and UTF-8 would yield not equal?
anyone doing a machine diff should be doing a machine diff of the raw
public key material in the first place, no?

The ssh "visual host key" artwork is for human consumption.  i'm
rather
dubious about it its security properties, i confess, but i don't think
there is any reasonable context in which it should be
machine-interpreted.

       --dkg
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 948 bytes
Desc: not available
URL:
<http://lists.mindrot.org/pipermail/openssh-unix-dev/attachments/20150708/7fbfda4a/attachment.bin>

Instead of proposing encoding changes of output that would require a
tool to sanitise the resulting from these experiments, why not consider
using a tool outside SSH code to produce the pretty print in the first
place?