On 27 jun 2004, at 22:37, Richard MAHONEY wrote:
On Sun, Jun 27, 2004 at 12:29:44PM +0200, Maarten Sneep wrote:
On 26 jun 2004, at 17:09, Maarten Sneep wrote:
Installing Xindy involves compiling and installing (gnu)CLISP,
combining the Xindy CLISP sources into a kernel (with the CLISP
engine) and creating a format. I tripped at step 1.
I've tried again, and as far as I can tell xindy is *broken*: it is in
a non-compiling state right now: you need to use an older clisp
version
in order to use some modules needed by xindy, but that version will
not
compile on Mac OS X. Newer versions of clisp do compile and pass the
tests, but the modules xindy uses expect a different API, and can
therefore not be included into the clisp engine.
I'm giving up on xindy, it is a mess. BTW: from what I can tell from
the documentation: xindy uses the same tex commands as Makeindex, so
exectly what you can do with xindy that is impossible with Makeindex
(apart from using it with other document source formats, but the
source
here is (la)tex anyway) is unclear to me. what you _can_ do with
Makeindex, and what is impossible with xindy is installing.
It's a mess.
Have you mailed your concerns to the maintainer ?
I've posted a request for help to the xindy mailing list, and Joachim
responded.
The binaries that are available were compiled in 2000 or thereabout. I
received the
version number of clisp they used, and I'm about to try that one. If
that fails,
I will vent onto that mailing list.
Maarten
--
The reply from the XindY mailing list:
MS> I've been trying to get Xindy to run on Mac OS X. This means
MS> compiling clisp, and integrating the modules. As is explained on
MS> your web-page: this is impossible right now, as some of the
MS> required modules expect a different API than is offered by the
MS> current clisp.
MS> Does that mean that the build of xindy is broken? Could I try an
MS> older clisp release?
On Linux and Solaris, CLISP 2.24 is most probably used.
Actually, I really don't know how hard it will be. I know that CLISP
2.31 introduced a new foreign function interface (FFI), and afterwards
reports dropped in that one cannot build the xindy run time engine
(RTE) from source. I only had a brief look and don't know where it is
really broken -- please note that my cursory look didn't reveal
obvious problems, the used CLISP FFI declarations were still
documented. In fact, they are still documented in current CLISP
versions; I just spend an hour browsing the documentation there and
checking the current state of affairs.
Since I didn't want to delve in that problem, I just reported the
issue on the Web site. Perhaps you can elaborate what really goes
wrong when you try to follow Thomas Henlich's build report, using
CLISP 2.30? Compilation errors? Link errors?
MS> How do _you_ create the binary releases?
I create them by compiling the xindy source with the available RTE,
and add the xindy modules. My current work is focused on the
ready-to-run end-user issue and on working xindy modules, not on new
RTEs. This will change when I have sorted out the current issues and
will work on a Windows port and better Unicode support, but I'm not
quite ready for that yet.
I think Thomas Henlich created the current RTEs in 2000, the one's
that I'm using now. His report how he created them is on the Web site.
Perhaps he may elaborate which CLISP version he was using -- as I
said, I suspect it's 2.24 since the output of the
LISP-IMPLEMENTATION-VERSION function fits to that release.
Cheers,
Joachim
| 