Re: [Groff] Broken chars

[Werner LEMBERG]

> >The .tr request translates characters.  In this particular case, it
> >translates `|' to `\(bv'.  `bv' is equivalent to `braceex' in PS
> 
> Hm, here's a suse'd perlcheat.1.gz, which when viewed with a plain text 
> editor, contains $|, and not \(bv.
> (http://jengelh.hopto.org/f/perlcheat.1.gz)

Of course!  And the line

  .tr ... |\(bv ...

in the preamble of the man page converts it to the `bv' entity in the
output.

> In the "devutf8/R" file is:
> ba      24      0       0x007C
> or      "
> |       "
> br      24      0       0x2502
> bv      "

Aah, sorry, I've changed the `bv' mapping in later groff versions
(U+2502 is wrong).

> >   `  the ISO Latin-1 `Grave Accent' (code 96) prints as <U+2018>,
> >      a left single quotation mark; the original character can be
> >      obtained with `\`'.
> 
> Obviously, noone uses \`.

I do.  But you are missing the point.  It's the job of pod2man to
provide the correct mapping, not the user, for example by calling

   .tr ... `\` ...

before handling verbatim text.

> >  .if '\*[.T]'utf8' \{\
> >  .  char \- \N'45'
> >  .  char  - \N'45'
> >  .  char  ' \N'39'
> >  .\}
> 
> Yes the andocdb is a hack, really.

Indeed.  It has been added only because most software isn't smart
enough to allow an easy search for all hyphen characters at the same
time with a single command.

> But adding all hacks, including the | mentioned above, would be an
> overkill.

No hack necessary!  Just fix pod2man to avoid this braindead mapping
(at least for groff).

> Well, I do not know why they are, I guess because they don't fit in
> any other category-
> 2-syscalls and stuff          3-libc functions
> 4-block and char devices      5-config files
> 6-games                               7-protocols
> 8-sysadmin                    9-kernel

I would put it into section 7.  It describes a `protocol', namely perl
input files look like.  For example, there is also `groff(7)' which
gives a short reference of the GNU language.


    Werner