New personal experimental book [warning: lots of UTF-8 in this]
dj at linuxfromscratch.org
Wed Oct 1 18:43:34 PDT 2008
Colin Watson wrote:
> On Tue, Sep 30, 2008 at 09:03:00PM -0500, DJ Lucas wrote:
>> Other packages in the base LFS utilize BDB. They may or may not work
>> with GDBM so I'll be looking into that as soon as we get updated to
>> reasonable revisions of all installed 'base' software. My question,
>> however, will man-db-2.5.3 allow continued used of BDB in the near future?
> Yes (--with-db=db, or --with-db=dbN for N=1-4), although I can't promise
> to test it very often.
> We use the notorious Debian-patched groff 18.104.22.168 and configure man-db
> with --enable-mb-groff. I'd rather that not be the only possible
> alternative, of course.
>> My real concern is the version of groff being used. I did not see
>> mention of a current groff version which was *my* original concern. I
>> want to use what works, but I also want to stay as close to upstream as
>> possible for all packages because we (LFS) do not have the development
>> staff that distributions have. Keep in mind that LFS is an educational
>> product, not a 'distribution', though many use it as their
>> 'distribution' of choice. Utilizing Debian's work in this area was
>> great (and will continue to be I think). It allowed Alexander to
>> provide a working setup for almost all cases, and explain in detail the
>> future issues (though the current text, like much of the book ATM, is
>> now out of date).
> For staying as close to groff upstream as possible, you probably want to
> use the preconv preprocessor included in CVS groff. That eliminates the
> need for the Debian multibyte patch for most languages. Unfortunately
> there has been no new upstream release of groff since that work was
> The remaining problem is that nobody's yet finished the work on
> character classes in groff, which mean that certain kinds of specialised
> typography don't work: in particular the line-breaking algorithm
> required for Japanese text ("kinsoku shori") isn't implemented. This is
> the reason we're still sticking with the multibyte patch in Debian for
> now, since I want to avoid introducing regressions. I think everything
> other than CJK should work with preconv, although feedback from people
> actually regularly using it wouldn't hurt.
Thanks for the detailed response Colin. For the immediate future, LFS
will have to stick with your known working method. Maybe later we can
look into backports of groff-cvs.
-- DJ Lucas
This message has been scanned for viruses and
dangerous content, and is believed to be clean.
More information about the lfs-dev