[blfs-dev] Success with xindy on i386

Ken Moffat zarniwhoop at ntlworld.com
Wed Jan 21 14:16:39 PST 2015

On Wed, Jan 21, 2015 at 03:48:43PM -0600, Douglas R. Reno wrote:
> If I have the chance (and I am extremely busy), I have about 12 i686
> machines in my house from a wide variety of ages (1995-2008) and I might
> try building a modern LFS/BLFS system on them and then try TexLive. It will
> take me a very long time, but I might be able to.
> Douglas R. Reno

 I will note that I never built xindy on i686 until {,B}LFS-7.6.  I
initially tried the 32-bit binary on the desktop which sometimes runs
32-bit (it lacks CPU horsepower and since we had the binary, I
wanted to try that somewhere - that is the box where the 64-bit
luatex binary was broken :-), but then decided to try a VM while
running the x86_64 system.  Actually, I never had it working on
x86_64 until last summer so I suppose this is still early days.

 But xindy (the area of the problem) is separate, and now built
after texlive.

 Iff you want to try it, my current TeX test scripts (which all
produce PDFs) are in my LFS webspace.  Taking a rough view, texlive
itself does not need a vast amount on top of _my_ normal desktop - I
suggest you begin by building whatever you use, but with an eye to
the recommended deps for texlive itself (or, use the versions
shipped in texlive if that is your preference).  Xindy itself is
quite a quick build - Pierre fixed the clisp tests, and clisp now
seems to build quickly.

 The slow part of texlive from source is downloading the texmf
tarball.  And it uses a *lot* of disk space (distros build most of
texlive, but then split the parts up into different packages - my
method builds it all because I cannot see any reliable way to limit
which parts of texmf get installed (e.g. files for xe{,la}tex are
spread throughout the texmf tree).

Nanny Ogg usually went to bed early. After all, she was an old lady.
Sometimes she went to bed as early as 6 a.m.

More information about the blfs-dev mailing list