A86: 86g File Quirks


[Prev][Next][Index][Thread]

A86: 86g File Quirks



I just spent the last couple hours tracking down an odd quirk with some
86g files.  I have had many files cause strange errors in xlink86 (my
personal hacked up version of xlink85 for use on my linux box) when I
tried to send them.  After ruling out xlink (lots of gdb), I noticed
there seems to be a recurring error in the files.  Referring to the
85g.txt (which should be the same for 86g) file:
<snip>
  +05h      01h       Byte: Name length
  +06h       *h       Name (plain data, no terminating null or padding
                      spaces)
<snip>

At the name length offsets in all of the files that didn't work, I found
that the length stored there was correct (0x05 for a variable named
"Sqrxz") but the name that followed always contained 8 characters.  In
the Sqrxz.86g, the Sqrxz was followed by 3 spaces (0x20) then the
correct data length.  In Aurora, the first variable, APic1, was followed
by null characters.  Later in the same file, the variable named AuroraO2
(which has 8 letters) has a name length of 6 stored in the file.  It
seems like whatever program that everyone else is using ignores the name
length byte and assumes that the name has 8 characters no matter what. 
Did I miss something in the standard?  Do 86g files differ from 85g
files?

Also, I was wondering if everyone uses the same program for converting
their object files to 85p/85g files?

Stan Seibert


Follow-Ups: