Subj : Re: spaces in filenames
To   : Oli
From : tenser
Date : Thu Jan 20 2022 04:52 am

On 19 Jan 2022 at 03:57p, Oli pondered and said...
 
 Ol> tenser wrote (2022-01-20):
 Ol> 
 Ol>  t> * The FSP-1011.003 escape method of "\HH" is deprecated.
 Ol> 
 Ol> It's not deprecated, it was never valid binkp. FSP-1011 was never
 Ol> released as a standard and \HH was a documentation error. (if my
 Ol> interpretation is correct)

I think you're splitting semantic hairs here.  There is clearly
software in the wild that implements the proposal's behavior;
that behavior has been defined to be "incorrect" in an actual
standard, but nonetheless legacy software implements it.  In
FTS-1026.003, this is described as "the incorrect method", but
given that the dictionary definition of "deprecate" as a verb
is to "express disapproval of" I'd say that 1026 certainly
deprecated the proposal's mechanism.  :-)

 Ol>  t> * Systems should provide a configuration option on a per-node
 Ol>  t>    basis to use the FSP-1011.003 method for nodes running software
 Ol>  t>    written assuming the "\HH" encoding.
 Ol> 
 Ol> "It is _advised_ to have a setting for specific nodes to sent escaped
 Ol> characters using the incorrect method."
 Ol> 
 Ol> advise is not a SHOULD or MUST.

Again, splitting semantic hairs.  I did not say I was using
the RFC terminology; my use was intended to be colloquial.

 Ol> It was not the worst idea when the FTS was released, but it's not really
 Ol> a sufficient workaround and in many cases useless. In the source code and
 Ol> documentation of several mailers, only Argus and Irex are mentioned as
 Ol> the mailers that are victims of the error in FPS-1011. I'm not sure
 Ol> anyone is using Argus anymore (and it has became open source, so it
 Ol> could be fixed). And Irex binkp implementation is a mess anyway and
 Ol> might need some additional workarounds
 Ol> (https://gitlab.synchro.net/main/sbbs/-/issues/185).

I agree with you.

 Ol> All other mailers I know do use \xHH or don't do any escaping at all.

If they do no escaping, they are wrong.  The standard is quite
clear about this.

 Ol> IMHO, don't waste time with implementing and documenting a configuration
 Ol> option. Do some hard coded auto detection for the few specific mailer
 Ol> versions that have problems or just ignore the few broken mailers.

This begs the question: how does one define a "configuration
option"?  That could be a dynamic determination made when the
connection is established based on some heuristic.

FTS-1026.003 is very poorly written; at a minimum, it needs a
good editor to clean up the grammar and language.  My suspicion
is that the "configuration option" bit wasn't meant to be
strictly literal, but rather, that the intent be some mechanism
for dealing with software written to the proposal's recommendation,
as a pragmatic consideration.  Whether that was done with some
file or done with some other mechanism seems largely irrelevant
from the intent.

 Ol>  t> * Software should unconditionally transform "\20" (specifically)
 Ol>  t>    into a space character in file names.
 Ol> 
 Ol> .... and send a warning message back ;)
 Ol> 
 Ol> M_NUL BUG "Your mailer is a victim of FSP-1011. Use \x20 instead of \20."

Let's be honest: most BBS people would have no idea what to do
with that and would just ignore it.

--- Mystic BBS v1.12 A47 2021/11/06 (Linux/64)
 * Origin: Agency BBS | Dunedin, New Zealand | agency.bbs.nz (21:1/101)

.