969dad87eb
fix a secfault in return_expr
2003-10-22 08:30:54 +00:00
72ae59f882
short circuit logic can now be controlled (see man page)
2003-10-22 08:27:38 +00:00
945006494f
short circuit logic for && and || (as per C). forces standard precedence
...
(ie, breaks --traditional) when enabled (no choice at the moment, next fix)
2003-10-22 08:05:17 +00:00
99034fef27
initialized locals aren't constant
2003-09-25 02:05:43 +00:00
4ce81b28d7
remove nested scopes for traditional mode
2003-09-23 17:59:46 +00:00
91207c0180
make void functions actually return float in traditional mode (qcc allowed
...
things like foo = void_function)
2003-09-23 16:57:53 +00:00
387464a27c
mangle precedence of = relative to && and || for --traditional (bletch).
...
also make noise when assigning to bad lvalues (eg !x = y)
2003-09-23 16:42:03 +00:00
ba26628b80
--traditional now "fixes" (ie, breaks) operator precedence to match qcc
...
except for the relationship between "=" and "&&" and "||": not sure I want
to go that far towards bug compatability.
2003-09-22 22:15:31 +00:00
9027648721
kill -t for --traditional, add --advanced, default to --traditional for
...
progs.src mode and --advanced otherwise
2003-09-20 04:13:32 +00:00
651fa8ca12
correct the operator precendece to match C. hopefully won't cause any
...
problems :)
2003-09-19 20:50:05 +00:00
11538d4745
access arrays in structs more efficiently (especially when using constant
...
indeces)
2003-09-02 17:41:16 +00:00
f10829206a
use a cast expression rather than and bind expression to do the unsigned
...
comparison for switches. clears the notice (bind seems to be correct in
general) and is more correct conceptually anyway.
2003-09-02 16:53:45 +00:00
23b525059f
cast between integer and unsigned just by changing access method
...
effectively deprecating conv.ui and conv.iu (they'll die eventually:)
2003-09-02 16:45:36 +00:00
3a6c8b8b16
wrong place to decrement uses. now have a problem in switch :/
2003-09-02 06:23:57 +00:00
ef4ca0c798
some doxygen tweaks
2003-08-31 05:24:24 +00:00
b01d4d4589
get the type for the destination from the right place (hope that wasn't
...
a bugfix for something else:)
2003-08-24 20:38:32 +00:00
7cffeb27bf
correct the temp def counting when casting
2003-08-24 08:17:31 +00:00
6086aa2b99
allow casts between pointers and integers
2003-08-24 08:03:25 +00:00
a08efafb7a
make sure __VERSION6__ is defined when it should be
2003-08-24 07:26:02 +00:00
385a90e8fb
correct pr_symtab_t to include the pointer to the array of selectors used
...
in the module. Unfortunatly, this requires a version bump on both qfo and
progs files due to the changes structure layout.
2003-08-23 06:15:19 +00:00
b38486d405
make rel_def_op work properly
2003-08-22 19:48:14 +00:00
e9ca03416a
create the temp def for bound defs properly
2003-08-22 18:16:30 +00:00
d20b8bf0c1
pass NIL though ... safely
2003-08-22 06:04:31 +00:00
303287dc11
don't seg on undefined vars
2003-08-22 05:48:19 +00:00
204243fe95
better message type checking
2003-08-22 05:26:47 +00:00
febc57e71b
typecheck message params when possible
2003-08-22 04:08:33 +00:00
7d09105740
remove debug code
2003-08-21 02:24:11 +00:00
11c7c27f70
better relocations: pointers to structure members now work
2003-08-21 02:21:30 +00:00
a2300387de
fix a bunch of segfaults
2003-08-20 22:36:02 +00:00
93cafe878a
convert NIL properly in function calls
2003-08-20 20:28:28 +00:00
7d830ffc1f
make @defs work properly and don't do uninitialized checks on structs
2003-08-20 19:58:41 +00:00
aa712c0169
make [super foo] work in category methods
2003-08-18 06:11:41 +00:00
aa4f81b6da
class.c:
...
whitespace
switch.c:
set up the relocations for the switch jump table properly
2003-08-18 02:57:21 +00:00
515dbaccc0
make qc-lex.c and qc-parse.[ch] part of the distribution (two less
...
requirements for the average user)
2003-08-12 18:22:11 +00:00
4a680bf981
grovel around for the source file and line of an undefined def to give
...
better diagnostics
2003-08-06 21:36:07 +00:00
448b6f24be
make the useless expression warning more interesting
2003-08-05 21:41:48 +00:00
329c91fce4
expr.c:
...
qcc actually had the precedence of "!x && y" correct, so don't invert
the logic for it in traditional code.
qfcc.c:
allow the .debug_file def to be emitted in traditional mode.
2003-08-05 17:48:16 +00:00
3e3b316891
blah, double ;
2003-08-01 21:26:36 +00:00
dd9ab1e050
get unsigned mostly working
2003-08-01 21:20:04 +00:00
8e21ecd4d9
make private methods work properly and put overided methods into the
...
correct class
2003-08-01 16:24:44 +00:00
1b5bf9d176
make methods showing up only in the implementation work when overiding an
...
inherited method
2003-08-01 05:08:15 +00:00
24f4bcfebf
initialize the new field properly
2003-08-01 02:43:51 +00:00
0361cb261f
fix an incorrect inherited attribute access caused by my "no ivars" fix
...
(part of the struct re-work)
2003-08-01 02:43:11 +00:00
ae8e7553ce
fix struct parsing and printing
2003-07-31 06:17:49 +00:00
9f0d4cc746
must use save_string when source of string is unknown :P
2003-07-31 06:14:26 +00:00
cc0d8b364d
fix a couple "value computed is not used" not used warnings
2003-07-31 02:16:26 +00:00
ab94498503
heh, that was a symptom. forget &void :)
2003-07-30 21:55:32 +00:00
8099f0d8f2
handle addresses of void defs
2003-07-30 20:57:50 +00:00
e957dc798f
oops, initialize type_param properly (D'OH:)
2003-07-30 17:59:34 +00:00
6ecb707257
revamp the struct system so "struct foo;" does the right thing (ala C)
2003-07-30 04:11:45 +00:00
Bill Currie