These notes will help people wanting to use ISO 8859-1 stuff in
WAIS.  We have only done this on Sun's running SunOS 4.1.3. To
run it successfully under SunOS 4.1.1, one will need patches
from Sun to correct problems with COLLDEF(8). 

1- Install the collation table.

You might already have one, check /etc/locale/LC_COLLATE to see
what's in it. You probably have only C and default in there.  
Due to a bug in the SunOS 4.1.3, it is useless to have the thing 
in /usr/share/lib/locale/LC_COLLATE since it's ignored. (This
caused us much grief.)

We decided to use external definitions of the 8859-1 table 
which we also put in /etc/locale/LC_COLLATE; we called this
file iso_charmap but you can use anything as long as the same
name appears in the input to COLLDEF(8).

As root, 
/usr/etc/colldef <your_input_file /etc/locale/LC_COLLATE/iso_8859_1
will install the table. Our copies of these files can be found at 
the end of this document. By the way, my version of colldef gives
a checksum of 02951  24.

One is not obliged to use the name iso_8859_1, but (for us at least)
it tends to keep it simple since this is exactly what we are doing!
If you do use something else, the lines containing
    setlocale(LC_CTYPE,"iso_8859_1");
    setlocale(LC_COLLATE,"iso_8859_1");
will have to be changed to whatever is needed.


2- Recompile and reinstall WAIS.

This part is easy. Just add -DLOCALE to the CFLAGS line in the
primary Makefile and make.  What this does is replace every call
to STRCMP with a call to STRCOLL instead. It also adds calls to
setlocale in the main programs.

The only other point worth mentionning: even if we stuck in
    #include <locale.h>
    setlocale(LC_CTYPE,"iso_8859_1");
    setlocale(LC_COLLATE,"iso_8859_1");
in the source, THIS DOES NOT SUFFICE! One must also have
   LC_CTYPE=iso_8859_1
   LC_COLLATE=iso_8859_1
in the run-time environment. This also caused us much grief.
This problem might be specific to Suns under 4.1.3

Note that Sun has put in the right stuff for LC_CTYPE which
they have not done for LC_COLLATE.

One might also want to add to the stoplist. To more or less
quote from the docs, when one generates an index, the
   -stop stop_list_file_name
option permits the use of another list.
We've called our (French ISO 8859) stoplist stop.uqam and
put it in here.

You might want to add the setlocale stuff to xwais if you're
using it. In ~/x/xwais.c and ~/x/xwaisq.c, add
    #include <locale.h>
    setlocale(LC_CTYPE,"iso_8859_1");
    setlocale(LC_COLLATE,"iso_8859_1");

3- Gopher stuff.

If WAIS is called by Gopher, gopherd must be changed for all this
to work. The same 3 lines must be added to ~/gopherd/gopherd.c, ie
    #include <locale.h>
    setlocale(LC_CTYPE,"iso_8859_1");
    setlocale(LC_COLLATE,"iso_8859_1");
and the run-time environment must contain
   LC_CTYPE=iso_8859_1
   LC_COLLATE=iso_8859_1
for the same reasons as in WAIS.


Good luck!

Sylvie St-Georges st-georges.sylvie@uqam.ca

Our iso_charmap file
--------------------------------------------------
A-grave		\xc0
A-acute		\xc1
A-circu		\xc2
A-tilde		\xc3
A-diaer		\xc4
A-ring		\xc5
AE		\xc6
C-cedil		\xc7
E-grave		\xc8
E-acute		\xc9
E-circu		\xca
E-diaer		\xcb
I-grave		\xcc
I-acute		\xcd
I-circu		\xce
I-diaer		\xcf
ETH		\xd0
N-tilde		\xd1
O-grave		\xd2
O-acute		\xd3
O-circu		\xd4
O-tilde		\xd5
O-diaer		\xd6
MULT		\xd7
O-stroke	\xd8
U-grave		\xd9
U-acute		\xda
U-circu		\xdb
U-diaer		\xdc
Y-acute		\xdd
THORN		\xde
s-sharp		\xdf
a-grave		\xe0
a-acute		\xe1
a-circu		\xe2
a-tilde		\xe3
a-diaer		\xe4
a-ring		\xe5
ae		\xe6
c-cedil		\xe7
e-grave		\xe8
e-acute		\xe9
e-circu		\xea
e-diaer		\xeb
i-grave		\xec
i-acute		\xed
i-circu		\xee
i-diaer		\xef
eth		\xf0
n-tilde		\xf1
o-grave		\xf2
o-acute		\xf3
o-circu		\xf4
o-tilde		\xf5
o-diaer		\xf6
DIVIS		\xf7
o-stroke	\xf8
u-grave		\xf9
u-acute		\xfa
u-circu		\xfb
u-diaer		\xfc
y-acute		\xfd
thorn		\xfe
y-diaer		\xff
--------------------------------------------------

Our input file to colldef
--------------------------------------------------

charmap /etc/locale/LC_COLLATE/iso_charmap

substitute "\xc6" with "AE"
substitute "\xdf" with "ss"
substitute "\xe6" with "ae"

order \x00;...;\x20;\x21;\x22;\x23;\x24;\x25;\x26;\x27;\x28;\x29;\
	\x2A;\x2B;\x2C;\x2D;\x2E;\x2F;0;1;2;3;4;5;6;7;8;9;\
	\x3A;\x3B;\x3C;\x3D;\x3E;\x3F;\x40;\
	\x5B;\x5C;\x5D;\x5E;\x5F;\x60;\x7B;\x7C;\x7D;\x7E;\x7F;\
	(A,<A-grave>,<A-acute>,<A-circu>,<A-tilde>,<A-diaer>,<A-ring>,\
	 a,<a-grave>,<a-acute>,<a-circu>,<a-tilde>,<a-diaer>,<a-ring>);\
	(B,b);(C,<C-cedil>,c,<c-cedil>);(D,d);\
	(E,<E-grave>,<E-acute>,<E-circu>,<E-diaer>,\
	 e,<e-grave>,<e-acute>,<e-circu>,<e-diaer>);\
	(F,f);(G,g);(H,h);\
	(I,<I-grave>,<I-acute>,<I-circu>,<I-diaer>,\
	 i,<i-grave>,<i-acute>,<i-circu>,<i-diaer>);\
	(J,j);(K,k);(L,l);(M,m);(N,<N-tilde>,n,<n-tilde>);\
	(O,<O-grave>,<O-acute>,<O-circu>,<O-tilde>,<O-diaer>,<O-stroke>,\
	 o,<o-grave>,<o-acute>,<o-circu>,<o-tilde>,<o-diaer>,<o-stroke>);\
	(P,p);(Q,q);(R,r);(S,s);(T,t);\
	(U,<U-grave>,<U-acute>,<U-circu>,<U-diaer>,\
	 u,<u-grave>,<u-acute>,<u-circu>,<u-diaer>);\
	(V,v);(W,w);(X,x);(Y,<Y-acute>,y,<y-acute>,<y-diaer>);(Z,z)
--------------------------------------------------

--------------------------------------------------------------------------
Jean-Pierre Kuypers <Kuypers@sri.ucl.ac.BE> writes the following:

About the credits, we'll not forget the works of Sylvie St-Georges
(Bonjour, Sylvie) for a ISO 8859-1'able version of WAIS. Pascal Maes uses
it to make his proposal. The work is available on ftp.uqam.ca:/pub/WAIS/,
where the files alire (French) and README.iso8859 (English) explain what
and how to do.

I try to install the new jughead version, on a SunOS 4.1.1. After some
troubles, as usual, I have now a jughead server able to search non-ASCII
items.

After uncommenting the CFLAG in the Makefile and making/installing, I must
do a lot of things before it works.

- I run the "colldef" command with the two Sylvie's files. But I must
delete all the "-", "<", and ">" is these files to avoid "Syntax error"
messages with colldef.

- I must set the LC_COLLATE environment variable (I had already the
LC_CTYPE) to iso_8859_1.

- I must do a link /etc/locale -> /usr/share/lib/locale. So, I don't put
the stuff in /etc/locale/LC_COLLATE/ (as writed by Sylvie), but in
/usr/share/lib/locale/LC_COLLATE/. 

- I must do "jughead -tB" to rebuild the correct index table. It's mandatory!

After that, I have a +/- good jughead server.
Curiously, e' (e-acute) and e` (e-grave) are seen as t and l. So, re'seau
and re`gles match as rtseau and rlgles. But they match correclty! To search
"re'seau", I may give "re'seau" or "rtseau". With other letters, as a`, u`,
e^, i^, o^, there is no problem. But E^ doesn't work.