Re: [Groff] Line-drawing escape

[Werner LEMBERG]

> What a scholarly expose!  This should be part of the manual.

Mhmm, patches welcome...  Such a detailed explanation is rather
something for the Wiki...


    Werner

> >> Your solution and Tadziu's are rather similar and both are working
> >> well:
> >>
> >> .de iS
> >> .char @ \0
> >> \\$1  \\kx\\l'(\\n(.lu-\\nxu-\w'\\$2  'u)@'  \\$2
> >> .char @
> >> ..
> >>
> >> .de iS
> >> .tr ~
> >> \\$1  \\kx\\l'(\\n(.lu-\\nxu-\w'\\$2  'u)~'  \\$2
> >> .tr ~~
> >> ..
> >>
> >> I hope that `char @' ends watching for @, and `tr ~~' ends watching 
> >> for
> >> ~.
> >> Gentlemen, thank you both.
> >>
> >> Miklos
> >
> > Apologies, Miklos! I should have added ".rchar @":
> >
> >   .de iS
> >   .char @ \0
> >   \\$1  \\kx\\l'(\\n(.lu-\\nxu-\w'\\$2  'u)@'  \\$2
> >   .rchar @
> >   ..
> >
> > The interpretation is that ".char xxx yyy" defines a character
> > (with metrics) whose name is "xxx" (in this case "@"), according
> > to the groff text "yyy" (in this case a simple "\0"). If "xxx"
> > is already the name of a character (as "@" is) then this is replaced.
> >
> > Then ".rchar xxx" removes that definition of "xxx" and (in general)
> > the name "xxx" disappears. However, if "xxx" (e.g. "@") is a
> > name which occurs in a font file, then groff will find it there
> > (you cannot remove "@" altogether since it is a name in a fint
> > file).
> >
> > Hence after
> >
> >   .rchar @
> >
> > you again have the normal interpretation of "@".
> >
> > Unfortunately, the use of ".char" does not "stack", so you can
> > only redefine a name once (it is like the use of "\fB ... \fP" in
> > this respect.
> >
> > If (for instance) you did
> >
> >   .char @ \0
> >      ....
> >   .char @ \(dd
> >      ....
> >   .rchar @
> >   .rchar @
> >
> > then the first ".rchar" would restore the name "@" to mean the
> > usual @ (i.e. it would drop back to @), and the second would
> > do nothing (since you cannot remove the name of a "normal" character,
> > i.e. named in a font).
> >
> > For example:
> >
> >
> > .char @ \0
> > .sp
> > address@hidden@address@hidden
> > .char @ \(dd
> > .sp
> > address@hidden@address@hidden
> > .rchar @
> > .sp
> > address@hidden@address@hidden
> > .rchar @
> > .sp
> > address@hidden@address@hidden
> >
> >
> > and compare this with:
> >
> >
> > .char \[xy] \0
> > .sp
> > This\[xy]is\[xy]some\[xy]text
> > .char \[xy] \(dd
> > .sp
> > This\[xy]is\[xy]some\[xy]text
> > .rchar \[xy]
> > .sp
> > This\[xy]is\[xy]some\[xy]text
> > .rchar \[xy]
> > .sp
> > This\[xy]is\[xy]some\[xy]text
> >
> >
> > In the first case you have assigned "\0" to the name "@",
> > and then assigned "\(dd" to the name "@", and then you
> > have removed the name "@" with the first ".rchar". But
> > the name "@" already exists in the standard fonts, so it
> > can be found again once you have removed it.
> >
> > In the second case, you have assigned "\0" to the name "\[xy]"
> > which does not exist anywhere else, and then assigned "\(dd"
> > to the name "[xy]", and then you remove the name "\[xy]"
> > and it vanishes forever.
> >
> > To come back to your own example above (using .char @), consider:
> >
> > .de iS
> > .char @ \0
> > \\$1  \\kx\\l'(\\n(.lu-\\nxu-\w'\\$2  'u)@'  \\$2
> > .char @
> > ..
> > .sp
> > .iS Text1 Text2
> > .sp
> > address@hidden address@hidden
> >
> > and you will see that ".char @" has not restored the normal "@"
> > nut has given the name "@" yet another meaning! The "true way"
> > is to use ".rchar"