> i agree (with kristaps).
> 
>> What we should do is ignore `Bk' and follow an easy algorithm for 
>> SYNOPSIS mode regarding breaking between children.  This, I think, is 
>> easier for us and manual writers alike.
>>
> 
> absolutely! SYNOPSIS is a hugely important part of the page - it's the
> first thing people see. and we don;t want it looking like shit. and we
> don;t want man page writers fannying around trying to make it look ok.
> it should just work sanely.
> 
>> Is `Bk' used often outside of this where .nr nS 1 is otherwise 
>> unacceptable (like those OpenBSD .9 manuals)?
>>
> 
> it is a synopsis-only thing. obviously we sometimes have mini synopses
> outside the main SYNOPSIS section. in those cases we could currently
> find Bk. i'm not aware of any other places it would normally occur.

Ingo and Jason,

I propose the following.  I've enclosed a patch that demonstrates this.

First, I disabled `Bk' completely.

Second, when `Op' or `Oo' is encountered with MDOC_SYNPRETTY (i.e., in a
SYNOPSIS section or with ".nr nS 1"), spaces are made non-breaking.

Example (mandoc -Owidth=70 ssh.1):

SYNOPSIS
     ssh [-1246AaCfgKkMNnqsTtVvXxYy] [-b bind_address]
         [-c cipher_spec] [-D [bind_address:]port] [-e escape_char]
         [-F configfile] [-I pkcs11] [-i identity_file]
         [-L [bind_address:]port:host:hostport] [-l login_name]
         [-m mac_spec] [-O ctl_cmd] [-o option] [-p port]
         [-R [bind_address:]port:host:hostport] [-S ctl_path]
         [-W host:port] [-w local_tun[:remote_tun]] [user@]hostname
         [command]

Note: this is after removing the escaped spaces that the manual author
had added here and there.

Example (mandoc -Owidth=70 spamd.8):
     spamd [-45bdv] [-B maxblack] [-c maxcon]
           [-G passtime:greyexp:whiteexp] [-h hostname] [-l address]
           [-M address] [-n name] [-p port] [-S secs] [-s secs]
           [-w window] [-Y synctarget] [-y synclisten]

This removes the `Bk' mess and gives a nice, regular algorithm for
formatting in MDOC_SYNPRETTY.

It can/will be trivially extended to the function macros, Fa and Fn and
family.

Thoughts?

Kristaps