Conference bulova::decw_jan-89_to_nov-90

    To my knowledge there are no generally available fonts other than
    Latin-1, DECtech, and Symbol at this time.  Particular geographies may
    have obtained and added fonts appropriate to their local languages, but
    they are not standard parts of DECwindows. 

The charset_registry and charset_encoding fields for ISO Latin-8 should be
ISO8859-8, not ISO-8859-8.

What Marty says is true, the base DECwindows font set only includes Latin-1
character sets, plus Symbol (for Mathematics), plus DECtech (for DECterm
line drawing and other stuff).  It is normally expected that each country
will add the appropriate fonts to their DECwindows kits, in addition to
doing things like translating all the help files and other text.

Maybe someday we will ship fonts for all languages as a part of the base
DECwindows kit, but I doubt that will be praticable until we have server
support for outline fonts instead of bitmap fonts.  Even with outline
fonts, I doubt we would ship Asian fonts as part of the base kit, unless
VMS switches to using CDROM exclusively for software distribution.

While these are not available as part of the base kit, they are available.
By contacting the people responsible for the localized kits, fonts with
the appropriate charsets can be obtained.  In some cases it pays to get
the whole localized kit (with tranlated .UIDs, etc.).

Dick

Does anyone have any contacts or know of anyone that can get the ISO 8859-6
font?  My customer has a large application that uses it and has upgraded to
vms 5.3 and no longer can run his applications.  Even better, does anyone know
how to find out the file name for iso 8859-6.  We tried searching his font 
directories for the ascii string "8859-6" and came up with nothing.  This
was on his vaxstation where is application still works because he hasn't
upgraded.  The file cda$def.h was the only place where I could fine a "link"
between the latin-6 font and the iso name.  Is this information reliable or
should I be searching for another string name?

ISO 8859-6 is the ISO Latin-Arabic Alphabet.

There is no official way with VMS V5.3 to get the font name that
corresponds with a particular font file, as only the server knows
how to interpret the format of font files.  However, the font
name is simply an ASCII string embedded in the binary font file,
so you should be able to dump the font file and see the font name
in the file.  I would expect the string "8859-6" in the name.

Here is another trick you could try: you say that your customer
has a VAXstation where the application still works.  Have the
customer run his application such that he is sure that the font
file is opened.  Then, with ANALYZE/SYSTEM, look at the server
process and do a SHOW PROCESS/CHANNELS.  You should see all
the font files in use by the server.  By a process of elimination
you should be able to figure out which one is the Latin-Arabic file.

Where the heck did this customer get the font files from in the
first place?  I would start asking there as to why things aren't
working!

Ok, this problem gets more interesting all the time.
I did what ya suggested (show proc/chan in sda)
and found that the following fonts were being 
opened by the server when the application was running.

Fonts opened under version 1 of decwindows:

DUA0:[SYSCOMMON.SYSFONT.DECW.75DPI]MENU12.DECW$FONT;1
DUA0:[SYSCOMMON.SYSFONT.DECW.75DPI]FIXED.DECW$FONT;1
DUA0:[SYSCOMMON.SYSFONT.DECW.CURSOR16]DECW$CURSOR.DECW$FONT;1
DUA0:[SYSCOMMON.SYSFONT.DECW.75DPI]TERMINAL14.DECW$FONT;1
DUA0:[SYSCOMMON.SYSFONT.DECW.75DPI]TERMINAL18.DECW$FONT;1

fonts opened under version 2 of decwindows:

PEACHS$DUA100:[SYSCOMMON.SYSFONT.DECW.75DPI]MENU12.DECW$FONT;1
PEACHS$DUA100:[SYSCOMMON.SYSFONT.DECW.75DPI]FIXED.DECW$FONT;1
PEACHS$DUA100:[SYSCOMMON.SYSFONT.DECW.CURSOR16]DECW$CURSOR.DECW$FONT;1
PEACHS$DUA100:[SYSCOMMON.SYSFONT.DECW.75DPI]TERMINAL_DECTECH14.DECW$FONT;1
PEACHS$DUA100:[SYSCOMMON.SYSFONT.DECW.75DPI]TERMINAL14.DECW$FONT;1
PEACHS$DUA100:[SYSCOMMON.SYSFONT.DECW.75DPI]TERMINAL_BOLD14.DECW$FONT;1
PEACHS$DUA100:[SYSCOMMON.SYSFONT.DECW.75DPI]TERMINAL28.DECW$FONT;1
PEACHS$DUA100:[SYSCOMMON.SYSFONT.DECW.75DPI]TERMINAL_NARROW14.DECW$FONT;1
PEACHS$DUA100:[SYSCOMMON.SYSFONT.DECW.75DPI]TERMINAL_BOLD_NARROW14.DECW$FONT;1


	Nothing real unusual there.  None of these fonts had a heading of
8859-6 in them.  So my search for this font was a waste of time.  
Apperently, somehow, the version one uil compiler was converting

character_set =  latin-6

into something else that it could use and was no longer doing this in
version 2 of the uil compiler.  To confirm this I did a

$uil/machine_code/list helloworld

on the version one uil file and then on the version two uil file.
I did a diff on the two listings and the following output is
of interest:

************
File PEACHS$DUB1:[PICKETT.C.DECW]HELLOWORLD.LIS;3
  151                                             0000   size: 69, group: literal 
  152                                             0000   type: compound string, access: private, locked: false 
  153       57440382 01018101 01800FA0 417FFFFF   0000   ��?A�?????????DW
  154       38384F53 4909810B 600DA100 C902A354   0010   T�?�?�?`???ISO88
  155       44000185 03B205A3 07621FA2 362D3935   0020   59-6�?b?�?�????D
  156       6F206E6F 74747562 20737365 72500012   0030   ??Press button o
  157                               00 4165636E   0040   nceA?
  158    
******
File PEACHS$DUB1:[PICKETT.C.DECW]HELLOWORLD_V1.LIST;1
  149                                             0000   size: 38, group: literal 
  150                                             0000   type: compound string, access: private, locked: false 
  151       73736572 50031383 80308030 8079FFFF   0000   ��y?0?0????Press
  152       00018100 65636E6F 206E6F74 74756220   0010    button once????
  153                             0000 00000000   0020   ??????
  154    
************

	The version two list shows that when it sees the latin-6 for the
default character set it puts in the string ISO8859-6.  But the version
one listing does not.  It puts in .... Nothing? (guessing).  
	Therefore version one works because it doesn't really look for
ISO 8859-6, but version 2 does.  I tried compiling the version 2 uil
file with the /version=v1 qualifier and still got the same results.
	My question now is, what font does version one use and how
can my customer get around this?

Something is fishy with your customer's problem, so let's go back
to square one to make sure we are all working under the same assumptions:

1.  Is your customer really using Arabic characters?  That is the only
    reason one would use ISO8859-6 instead of ISO8859-1.

2.  If so, does your customer really see Arabic characters on a
    VMS V5.3 workstation?  This is hard to believe, given what you've
    shown from SDA.  All of the files listed use Latin-1 characters,
    unless someone has switched modified font files for the standard
    distributed font files.

3.  Was the helloworld example that you gave modified by you or the
    customer to contain the string "latin-6"?  From the difference
    file, it looks like the text is still English.  This example
    would look the same using either Latin-1 or Latin-Arabic, since
    both character sets contain all the letters used in English.

I'm guessing that you or the customer modified helloworld.uil as an
example of a problem that the customer is having with a real application.
It would help things greatly if this was stated up front.

Someone else will have to answer questions about how different versions
of the UIL compiler deal with character sets.

Correctly using Arabic characters requires more than just a font change. The
application has to manage the context analysis on output (the shape of the 
character changes depending where it is in the string) as well as mixed
direction support.

Digital does have a version of DECwindows with Arabic support. If you are
interested contact me and I will find out who to talk to.

We also ship local versions supporting Japanese, Korean and Hebrew. Traditional
and Simplified Chinese is coming soon.

Tim