[tlbuild] xindy again

[Joachim Schrod]

On 12/15/15 20:44, Apostolos Syropoulos wrote:
> There is definitely something wrong withe either xindy or my xindy.
> I have two xindy "systems":
> 
> $ /opt/gnu/xindy/bin/xindy -V
> xindy release: unknown
> xindy script version: 1.08
> xindy kernel version: 2.3
> xindy run time engine: i386-solaris2.11, version 2.2
> CLISP version 2.43 (2007-11-18) (built on nadya [127.0.0.1])
>     architecture: I86PC
> $ xindy -V
> xindy release: 2.5.1
> xindy script version: 1.18
> xindy kernel version: 3.0
> CLISP version 2.48 (2009-07-28) (built 3478792388) (memory 3640346152)
>     architecture: I86PC
> 
> When I run the first on a .idx file I get
> [...]
> 
> When I run the new version I get
> [...]
> Reading raw-index "/tmp/QLpAj9duZ0"...
> WARNING: Would replace complete index key by empty string, ignoring
>          #<ordrule-regexp: '^ +' => '' :again NIL :only-at-start T>
> WARNING: Would replace complete index key by empty string, ignoring
>          #<ordrule-regexp: '^ +' => '' :again NIL :only-at-start T>
> WARNING: Would replace complete index key by empty string, ignoring
>          #<ordrule-regexp: '^ +' => '' :again NIL :only-at-start T>
> WARNING: Would replace complete index key by empty string, ignoring
>          #<ordrule-regexp: '^ +' => '' :again NIL :only-at-start T>
> WARNING: Would replace complete index key by empty string, ignoring
>          #<ordrule-regexp: '^ +' => '' :again NIL :only-at-start T>
> ERROR: CHAR: index 0 should be less than the length of the string
> 
> Is this normal?

Most probably: yes. The new version notifies about a problem you
had before, but maybe didn't notice.

The most probable cause is that you have index entries that consist
only of TeX macros that xindy doesn't know about. (That may also be
LICR representations of national characters that xindy doesn't know
about.) Then xindy replaces them with the empty string and merges
them all to one entry -- which ain't what you want.

Second on the probability list are merge rules used by you that
cause xindy to discard the whole index key. (Some highly rated
recommendations from tex.stackexchange.com are horribly wrong; the
authors obviously didn't understand xindy's basic functionality at
all.)

Especially the warning cited above is concerning. It is about the
rule that strips leading spaces. If that results in an empty index
entry, something is not right here. (That check was added sometimes
in the last 2 years, that's why you see it now and not before.) The
actual question to ask here is: Where is the rest of the index
entry? Why would stripping leading spaces result in an empty index
entry?

Sorry that I can't be of more help,

	Joachim

PS: Your log had lots of other errors. If you want me to analyze
more of your problem, I'll ask you to contribute a minimal example.

-- 
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Joachim Schrod, Roedermark, Germany
Email: jschrod at acm.org
xindy maintainer