Re: [Groff] Re: html.cc patch

[Gaius Mulley]

Werner writes:

> As you can see, you have sent me a reverted patch.  I have applied
> it.  A quick test with `man.1' still shows a lot of problems.  Never
> mind :-)

oops sorry about the reverted patch. Yes the man.1 is still causing
problems :-( I'll work on this and Lester's paper over the Christmas
vacation.

> What do you think about the following algorithm:

a fine algorithm IMHO !

>  0) The groff package will always use the `g' prefix.
great

>
>  1) Check whether a binary called `troff' is installed.  If yes, set
>     HAVE_TROFF.  Now check whether it is GNU troff.  If yes, set
>     HAVE_GTROFF_AS_TROFF.
>
>  2) Check whether a binary called `gtroff' is installed.  If yes
>     check whether the installation directories are the same.  If not,
>     emit a warning that another groff package is on the system which
>     can cause conflicts.  [Shall `make install' abort in this
>     situation?]

my vote would be 'yes' but perhaps we could include a 'make install-force'
target which ignores the presence of another groff.

>
>  3) Install groff.  Make a link gtroff->troff only if HAVE_TROFF
>     isn't set.  If HAVE_GTROFF_AS_TROFF is set, remove the old link
>     and use the old link's directory for the new link.  Do the same
>     for geqn, gtbl, etc.
>
> Are there any GNU guidelines covering such tricky situations?

I'm afraid I don't know of any GNU guidelines for these situations, thus
feel free to ignore my suggestions :-) It would be good if guidelines
were published and placed into the excellent info system in Emacs as per
"Managing Releases" "Makefile Conventions" etc

Well done Ted and Werner for coordinating the 1.16 release so smoothly,

cheers Gaius