xref: /csrg-svn/old/adb/common_source/CHANGES (revision 43900)
1*43900Sbosticde-Bourned; reformatted to Joy Normal Form
2*43900Sbostic
3*43900Sbosticold SDB relics clobbered
4*43900Sbostic
5*43900Sbosticlinted
6*43900Sbostic
7*43900Sbosticsome work done to prepare for /proc and discarding of ptrace
8*43900Sbosticinterface.  Reading sequences of bytes using ptrace is now done
9*43900Sbosticportably, and considerably more efficiently for commands that
10*43900Sbosticrepeatedly fetch objects smaller than an `int' (e.g., `l' command, `s'
11*43900Sbosticand `S' formats).
12*43900Sbostic
13*43900Sbosticadb used to accept 0t, 00t, 0<overflow-to-0>t, etc as decimal, and so
14*43900Sbosticforth.  now only the advertised sequences (and their uppercase
15*43900Sbosticcounterparts) are accepted: 0x, 0X, 0t, 0T, 0o, 0O.
16*43900Sbostic
17*43900Sbostic"unexpected ')'" error is gone; you now get "bad command".
18*43900Sbostic
19*43900Sbosticfixed crash if you asked for <expr> # 0
20*43900Sbostic
21*43900Sbosticfixed crash on input like 0.999999999999999999999999999999999999999
22*43900Sbostic
23*43900Sbosticsome expressions using variable 0 may behave differently.  In
24*43900Sbosticparticular, it used to be that `8000=C;<0=X' would first print ^@, then
25*43900Sbosticprint 8000.  Now the =C truncates, so that you get 0 instead of 8000.
26*43900Sbostic
27*43900Sbosticfinally got rid of that stupid
28*43900Sbostic	_foo:
29*43900Sbostic	_foo:	0
30*43900Sbosticoutput format.  _foo appears exactly once now.  (This may need some
31*43900Sbosticfurther work.)
32*43900Sbostic
33*43900Sbosticreenabled the 1, 2, and 4 formats (not that 2 and 4 are needed, as they
34*43900Sbosticare the same as w and W).  To get these you must write \1, \2, or \4,
35*43900Sbosticor the code will assume (as it used to) that these are prefix counts.
36*43900Sbostic
37*43900Sbosticthe VAX instruction decoder will now print comments in some questionable
38*43900Sbosticsituations that sometimes arise when attempting to print data as
39*43900Sbosticinstructions.  Previously it would print nonsense and rely upon the
40*43900Sbosticuser to discover it as nonsense.
41*43900Sbostic
42*43900Sbosticfixed radix code (previously everything that was not hex was decimal)
43*43900Sbostic
44*43900Sbosticexpression>pc now works as an alternate method for changing the pc
45*43900Sbostic(the usual expr:r, expr:s, expr:c formats still work).
46*43900Sbostic
47*43900Sbosticfixed annoying VAX feature that printed clrl/clrf instructions as
48*43900Sbosticclrf instructions, and likewise for mova and pusha: now l is preferred
49*43900Sbosticto f, q to d, etc., when two instructions have the same opcode.
50*43900Sbostic
51*43900Sbostic8a format no longer prints dot 9 times.  I doubt anyone will miss this...
52*43900Sbostic
53*43900Sbosticradix code now works more the old tahoe adb: a negative radix tells
54*43900Sbosticadb to print negative numbers with leading `-'; else all are assumed
55*43900Sbosticunsigned.  The new (internal and external) `v' and `V' formats print
56*43900Sbosticin the current radix, but always signed (used by, e.g., the vax
57*43900Sbosticinstruction disassembler, for fp offsets).
58