Message-ID: <323A17EF.1510@tempe.vlsi.com>
Date: Fri, 13 Sep 1996 19:26:55 -0700
From: Charles Marslett <charles DOT marslett AT tempe DOT vlsi DOT com>
Organization: VLSI Technology, Inc.
MIME-Version: 1.0
To: "John M. Aldrich" <fighteer AT cs DOT com>
CC: DJGPP Workers Mailing List <djgpp-workers AT delorie DOT com>
Subject: Re: "djverify" - DJGPP Installation Diagnostic Program
References: <323A1C3F DOT 3728 AT cs DOT com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

John M. Aldrich wrote:
> Wow!  In just one day I managed to pick up a fantastic amount of advice!
> Thank you, all.  :)  Now, on to some preliminary conclusions and additional
> questions:....
> 4) Initially, I think I will have it run in plain b/w text mode, with
> some basic windowing functions thrown in to make it look nicer.
> 
>    Question:  Am I safe in assuming that every computer the program
>    runs on will be capable of displaying the extended ASCII character
>    set (127-255)?  If not, is there any way to test for this?

I am fairly certain at least some systems sold in Europe do not have all
the PC line drawing characters (when I worked at STB two of my tasks were
to modify the fonts in our VGA adapters to implement the ISO-something-or-
other character set for sale in Denmark and Norway).  Double-byte countries
(Japan, Korea, China, etc.) probably do not show this characteristic because
of the size of the font compared to the space available for the BIOS ROM,
but our friends in Europe might want to comment on the prevalence of BIOSes
that support extended or ISO character sets.

As to testing for the characters, it is not rock solid, but you could examine
the font table pointed to by one of the vectors in low memory (EGA or VGA
only), and as a back up the second pointer to the upper half of the 8x8 font
table (not always present if you don't have an EGA or VGA, and not needed if
you do!).  YOu could look at the actual bitmaps used for the characters you
use, and from that ascertain if the font is usable or not.  Such a test
is not perfect, but it could be coded in a few lines, and would be reasonably
reliable.

I can email you a sample routine if you want.

--Charles