Adds for ALN - AL Brainstorm 1993-94.
(TABSIZE=4)
Main course: management of BSD and COFF encapsulated BSD file formats.
	Source level handled with DBX stabs.

5-may-1994
	? Moved gpu/dsp eqr test from ((add>=GPUEQR) && (add<=GPUEQR|31))
	  to (add&0xffffff80)==GPUEQR. (GPUEQR=0x80000080). Even if I don't
	  understand why.
	  Wish from Robert Dibley, Rebellion Software.
28-apr-1994
	+ Added a new display in case of unresolved symbols. If the '-u' option
	  is activated, the relocation locations are printed during the process,
	  together with source level informations if they are present. So,
	  aln -u .... gives something like:
	  Unresolved symbol: <mysymbol> ( in BSD object ) at offset 0X120.
	  Near /mywork/work/work.s:57.
18-apr-1994
	- Bug in this sh... BSD format in alignement. I "forgot" that BSD addresses
	  where TEXT absolute (not section relative!). So, data and bss relocations
	  and symbols were improperly aligned. MANY THANKS to ROBERT DIBLEY from 
	  REBELLION SOFTWARE! BTW, default phrase alignement is requested by Atari...
	- Bug in '-r's parameters recognition (was taking the following one)
	  (Robert, again).
	+ Added a warning when relocating with a GPU/DSP regequ value (Wish from ...
	  guess who?).
26-mar-1994
	+ Added '-r' option (as in MADMAC) which sets section alignment.
	  See addendum.aln for syntax. Unlike MADMAC, the default alignment
	  is PHRASE (8 bytes) for backward compatibility (the previous versions
	  of ALN for Jaguar were forcing BSS section size and included file sizes
	  to PHRASE alignment). Therefore, no more "Error: odd-sized TEXT or DATA
	  section; link aborted." and "Warning: DATA section size of xxx.o is
	  not phrase-aligned.".
21-mar-1994
	+ Added N_ASLINE stab (0x4c) which is a absolute source line. This is
	  an extension to standard DBX stabs.
10-mar-1994
	- Removed a bug in DRI symbol table check.
	! There seem to be problems under MS/DOS if you don't use the -B option.
26-feb-1994
	- Added '-ii' command line option, same as '-i' but doesn't truncate
	symbols to 8 cars.
19-feb-1994
	- Fixed a comparison bug in word fixups in BSD format on Atari
	(thanks to Rebellion Software's bug report - Hi Robert ;=)).
	- Moved long to unsigned long comparison for symbol name offset
	in BSD format (should only happen with crashed object files, thanks
	(again) to Robert).
 	- Fixed a print crash when not enough memory is available for the
	output COFF file.
1-feb-1994
	- Fixed common BSD labels bug (Thanks again to Loriciel's bug report).
31-jan-1994
	+ Common BSD labels are now handled (Thanks to Loriciel's bug report).
	- DRI output from BSD input is now ok (same as above ^^).
19-jan-1994
	- Bug in name offset for stabs removed.
	(RDB loads ok, get sources list with 'mlist' command.)
17-jan-1994
	+ Source level handling done. Have to test. Use '-g' option to get it.
	  Must be used with '-e' option. Not tested without '-a' option.
	  So you should have in your makefiles something like:
		For MAC:	-g -fb <source name>
		For ALN:	-g -l -e -a <text @> <data @> x -o <output name>
	  if you want source level.
	  (RDB doesn't load it yet anyway).
	! Careful: symbols generated with '-e' are now BSD! Please use
	  the next RDB version if you want to use it, even without source
	  level.
16-jan-1994
	- No more "unknown BSD symbol type" message (it should be stab!).
	? I don't understand how ld performs its xref pc relocation. I couldn't
	  make it work, and MAC and ALN work ok with my code. So it must be
	  a bug in ld's code or something... I don't think gcc generates
	  pc relative external references anyway so it could be kept hidden
	  until now.
	  Anyway, people should ALWAYS use ALN instead of ld.
14-jan-1994
	+ Output of COFF seems ok. The file extension is .abs, so you can use
		'aread <fname>'. But the file is not DRI at all ...
		RDB loads it now. So now you should have in your makefiles:
		For MAC:	-fb <source name>
		For ALN:	-l -e -a <text @> <data @> x -o <output name>
		With this you can have symbols of any length in link time,
		and with 23 cars max in RDB (RDB restriction only).
		For the moment, symbols are DRI (changed on 17-jan-1994 to BSD).
13-jan-1994
	+ Output COFF files (no source level yet). Use '-e' switch.
		(RDB doesn't load it yet anyway: done on 14-jan-1994).
11-jan-1994
	- Cleaned BSD object input. Now seems ok. Works with turtdemo.
6-jan-1994
	- Removed a bug in the dot check in output name
		(was misinterpreting something like ..\myprog)
5-jan-1994
	- Removed a bug that would not make a BSS absolute in a .abs file.
		It was caused by an extern int reference to a short value.
		(Don't know when it came).
	- When an empty line was typed in, its contents wasn't cleared.
		(This one seems to be there from the beginning).
4-jan-1994
	+ Added skipping multiply-defined locals (use -b to avoid it).
		This was made necessary because of the add of 'aread <absfile>' in RDB.
		One day I will make it more clever (yes it uses more memory :=)).
	+ Now takes BSD object files (Big endian only (Motorola), of course!).
31-dec-1993
	+ Added string table management.
	+ Symbols are now COFF-like, can read BSD objects.
	+ Relocation is function ptr'ed.
25-dec-1993
	+ Added bsdobj.c

