Follow-up Comment #4, bug #68719 (group groff): Hmmm, I think I want to strike out in a third direction altogether.
Check this out.
##$ printf 'included' > included
##$ cat included
included##$ cat sorceror
.so included
##$ soelim sorceror # GNU
.lf 1 "./sorceror
.lf 1 "./included
included
.lf 2 "./sorceror
##$ 9 soelim sorceror
.lf 1 "./sorceror
.lf 1 "./included
included
.lf 2 "./sorceror
##$ dwb soelim sorceror
.lf 1 "./sorceror
.lf 1 "./included
included
.lf 2 "./sorceror
##$ solaris10 soelim sorceror
included##$
##$ heirloom soelim sorceror
included##$
I think what we're seeing here is one of those rare beasts, a bug fix that
landed between DWB 2.0 (which, apparently, became the basis of "System V
_troff_" and thus went into Solaris) and DWB 3.3.
It wasn't very long ago that I learned why Heirloom Doctools has the name it
does: the name of the Solaris package that gets you "troff" is..."doctools".
I hadn't realized that all that "Heirloom" stuff was so strongly a Solaris
nostalgia trip. Myself, I'm a bit baffled why someone would be nostalgic for
Solaris. Sun's own customers resisted it mightily.
SunOS 4, _that_ was something I could see someone being nostalgic for, maybe.
You can see that if we're writing an `lf` request, **we have to have a newline
before it**. Otherwise it's meaningless and doesn't do its job.
But we also have an `-r` option.
-r Write files “raw”; do not add lf requests.
So I propose to:
* ensure that a newline ends an included file if we're not in "raw" mode; and
* don't bother adding a newline (but warn if one is absent) if we *are* in raw
mode.
_______________________________________________________
Reply to this item at:
<https://savannah.gnu.org/bugs/?68719>
_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/
signature.asc
Description: PGP signature
