1
0
mirror of https://github.com/golang/go synced 2024-11-18 00:54:45 -07:00
Commit Graph

147 Commits

Author SHA1 Message Date
Ian Lance Taylor
d12c1b99c7 Add ERROR comment for errmsg to look for.
R=gri
DELTA=1  (0 added, 0 deleted, 1 changed)
OCL=19460
CL=19500
2008-11-18 15:58:43 -08:00
Robert Griesemer
6c5fc055c8 import syntax incorrect
R=r
OCL=19457
CL=19457
2008-11-18 09:39:34 -08:00
Ian Lance Taylor
d8ecead73d The compiler should reject calling a const as though it were a
function.  Also update golden.out.

R=r
DELTA=18  (18 added, 0 deleted, 0 changed)
OCL=19433
CL=19448
2008-11-18 06:25:21 -08:00
Ian Lance Taylor
1945cc4c3c The compiler should reject comparisons between ints and nil.
R=gri
DELTA=8  (8 added, 0 deleted, 0 changed)
OCL=19434
CL=19436
2008-11-17 21:44:05 -08:00
Robert Griesemer
7692a93173 - new() accepts too many arguments
R=r
OCL=19413
CL=19413
2008-11-17 16:46:56 -08:00
Robert Griesemer
d58cd7626e - interface methods must have full function type
R=r
OCL=19410
CL=19410
2008-11-17 16:37:13 -08:00
Russ Cox
a1585b676b fix the easy parts of bug120
R=r,ken
DELTA=66  (52 added, 3 deleted, 11 changed)
OCL=19386
CL=19389
2008-11-17 13:58:45 -08:00
Russ Cox
f333f4685c floating point constant errors in 6g
R=r
OCL=19379
CL=19379
2008-11-17 12:33:49 -08:00
Ian Lance Taylor
b1e8b5f5b7 The scope rules have been clarified to indicate that a
variable may only be named after the complete declaration,
including the initialization statements.

R=gri
DELTA=61  (16 added, 45 deleted, 0 changed)
OCL=19343
CL=19376
2008-11-17 12:19:02 -08:00
Ian Lance Taylor
7dee51f491 The compilers should and do permit interface types to list
multiple method names with a single type.

R=r,gri
DELTA=10  (0 added, 10 deleted, 0 changed)
OCL=19292
CL=19298
2008-11-14 17:28:17 -08:00
Ian Lance Taylor
c4d8dc0b83 This bug has been fixed; moving it from bugs to fixedbugs.
R=gri
DELTA=36  (18 added, 18 deleted, 0 changed)
OCL=19059
CL=19133
2008-11-12 21:59:43 -08:00
Robert Griesemer
20b9bfb136 wrong code for array access
R=r
OCL=19107
CL=19109
2008-11-12 14:57:23 -08:00
Ian Lance Taylor
a957ceec35 The scope rules have been changed to say that labels live in a
separate per-function namespace.

R=gri
DELTA=24  (8 added, 16 deleted, 0 changed)
OCL=19006
CL=19057
2008-11-11 18:17:54 -08:00
Rob Pike
203a17c6ec update tests
R=gri
DELTA=28  (12 added, 14 deleted, 2 changed)
OCL=18923
CL=18923
2008-11-10 14:59:15 -08:00
Ian Lance Taylor
9c7374d71b Erroneous test case. The break statement should break out of
a select clause.

R=ken
DELTA=20  (0 added, 20 deleted, 0 changed)
OCL=18731
CL=18739
2008-11-06 17:39:48 -08:00
Ian Lance Taylor
ce15158502 Test that a break statement inside a select statement breaks
out of the enclosing loop.

R=ken
DELTA=20  (20 added, 0 deleted, 0 changed)
OCL=18686
CL=18714
2008-11-06 15:24:10 -08:00
Ian Lance Taylor
48d111f0b9 6g dumps core on this input file.
R=ken,rsc
DELTA=14  (14 added, 0 deleted, 0 changed)
OCL=18555
CL=18598
2008-11-05 15:27:40 -08:00
Ian Lance Taylor
7fe34ea002 Fix powser1.go to compile with the current 6g, which doesn't
recognize methods for a variable whose type is a named type
which is a pointer type.  Add bug117 to test this case.

R=r
DELTA=24  (22 added, 0 deleted, 2 changed)
OCL=18547
CL=18554
2008-11-05 12:06:48 -08:00
Russ Cox
56a7895386 fixed tests
R=r
DELTA=124  (62 added, 62 deleted, 0 changed)
OCL=18389
CL=18394
2008-11-03 15:52:34 -08:00
Rob Pike
c1198b9b87 update tests
R=rsc
DELTA=159  (65 added, 84 deleted, 10 changed)
OCL=18149
CL=18151
2008-10-30 13:26:57 -07:00
Russ Cox
cf9e46fec6 bugs from old gri mail
R=gri
OCL=18093
CL=18093
2008-10-29 19:34:47 -07:00
Ian Lance Taylor
e76e9cfc33 Call the right function for int32 values.
R=rsc
DELTA=2  (0 added, 0 deleted, 2 changed)
OCL=18073
CL=18073
2008-10-29 15:48:34 -07:00
Russ Cox
11a978adf5 shift bug in constant evaluator
R=ken
OCL=18031
CL=18031
2008-10-29 13:46:55 -07:00
Ian Lance Taylor
22a6747999 Per discussion earlier today with r and gri: when an interface
object has a value of type "int", it should not automatically
convert to type "int32".  That is, the type alias "int" should
be regarded as having been defined as though "type int int32"
appeared outside of the package, and as therefore being a
different type from "int32".

R=ken
DELTA=21  (20 added, 0 deleted, 1 changed)
OCL=17587
CL=17842
2008-10-24 22:40:32 -07:00
Rob Pike
4bcfffb761 update fixed bugs (and delete an irrelevant one)
R=rsc
DELTA=328  (158 added, 170 deleted, 0 changed)
OCL=17353
CL=17359
2008-10-17 11:48:51 -07:00
Rob Pike
13508ceabe bug is fixed
R=ken
OCL=17208
CL=17208
2008-10-15 11:49:34 -07:00
Rob Pike
151c0de8bc update bug list
R=ken
OCL=17169
CL=17169
2008-10-14 19:41:05 -07:00
Rob Pike
99c5b2ed4d 6g thinks a struct with one entry yields a composite that is a basic conversion
R=rsc
DELTA=12  (12 added, 0 deleted, 0 changed)
OCL=16989
CL=16998
2008-10-13 12:26:58 -07:00
Robert Griesemer
32d00e6cc6 - added missing ';' according to new rules
(no effect on the test, but now parseable)

R=r
OCL=16936
CL=16936
2008-10-10 16:18:01 -07:00
Russ Cox
e6cd1e445b remove uses of *T as an implicit forward declaration of T
R=gri,r
OCL=16648
CL=16652
2008-10-07 12:59:54 -07:00
Russ Cox
983f06bdb6 update code to follow new semicolon rules:
*  1. all statements and declarations are terminated by semicolons
 *  2. semicolons can be omitted at top level.
 *  3. semicolons can be omitted before and after the closing ) or }
 *	on a list of statements or declarations.

/home/rsc/bin/addsemi and then diff+tweak.

R=r,gri
OCL=16620
CL=16643
2008-10-07 12:31:31 -07:00
Russ Cox
74e2e087e2 change type names to go live at the name, so that
type T struct {
		next *T
	}

and

	type T *struct {
		next T
	}

are valid without needing forward declarations.

add "type T struct" syntax for forward struct declarations.

add "type T interface" syntax, but commented out
	(need to fix semicolons first)

R=ken
DELTA=452  (259 added, 115 deleted, 78 changed)
OCL=16580
CL=16584
2008-10-06 16:44:17 -07:00
Rob Pike
98c208ee1d bug026 and bug060 are fixed. update.
R=rsc
DELTA=87  (37 added, 50 deleted, 0 changed)
OCL=16498
CL=16507
2008-10-05 08:34:48 -07:00
Russ Cox
7766b12ead const bug, name bug - working on both
R=r
DELTA=43  (43 added, 0 deleted, 0 changed)
OCL=16468
CL=16475
2008-10-03 17:06:24 -07:00
Ian Lance Taylor
8d0ed065dc With the cleanup of method types, this test case is now
invalid.  The receiver type is not permitted to be a pointer
type.  Since 6g currently accepts it, this moves the test case
back to the bugs directory.

R=gri
DELTA=22  (11 added, 11 deleted, 0 changed)
OCL=16284
CL=16336
2008-10-01 17:01:42 -07:00
Ian Lance Taylor
aadd32223d Test that floating point constants resolve to type "float".
Currently 6g seems to resolve them to type "double".

R=ken,r
DELTA=30  (30 added, 0 deleted, 0 changed)
OCL=16132
CL=16152
2008-09-29 18:17:42 -07:00
Russ Cox
785afb0847 make bugs not fail
R=r
DELTA=6  (0 added, 0 deleted, 6 changed)
OCL=16110
CL=16114
2008-09-29 13:16:22 -07:00
Ian Lance Taylor
08df4dcf7d Storing an out of range constant into a variable should
ideally cause the compiler to give an error.  Right now 6g
warns about large shifts but does not give an error.  This CL
removes the out of range shift from shift.go, so that it will
work with gccgo, and adds a test case in bugs/ to be fixed at
a later date.

R=ken,r
DELTA=23  (9 added, 14 deleted, 0 changed)
OCL=16085
CL=16088
2008-09-29 11:03:21 -07:00
Ian Lance Taylor
6270e70b76 Add test case for 6g bug: result variable does not hide the
package name in the scope.

R=r
DELTA=12  (12 added, 0 deleted, 0 changed)
OCL=15913
CL=15926
2008-09-25 20:31:03 -07:00
Russ Cox
eeeaf6b931 6g misparses negative constants in imports
R=r
DELTA=8  (8 added, 0 deleted, 0 changed)
OCL=15893
CL=15898
2008-09-25 16:32:39 -07:00
Robert Griesemer
b6429768c6 bug: package identifier not visible
(should be in the scope chain like any other identifier)

R=r
DELTA=16  (16 added, 0 deleted, 0 changed)
OCL=15884
CL=15884
2008-09-25 15:45:19 -07:00
Ian Lance Taylor
b70d42f323 6g currently accepts setting a string value to an integer
constant without an explicit conversion.  I think that is a
bug.  This adds a test case for it.

Also, change errchk to include the string BUG in error
messages, so that failures are included in the count reported
by the run shell script.

R=r,ken
DELTA=11  (7 added, 0 deleted, 4 changed)
OCL=15857
CL=15880
2008-09-25 15:08:52 -07:00
Russ Cox
5383e28ea0 change string([]byte) to pass array, rather than &a[0],
to string convert.  if the byte array has length 0,
the computation of &a[0] throws an index bounds error.

for fixed size arrays, this ends up invoking arrays2d
unnecessarily, but it works.

R=ken
DELTA=304  (44 added, 28 deleted, 232 changed)
OCL=15674
CL=15678
2008-09-22 20:12:15 -07:00
Russ Cox
fb2c66710c compiler changes:
export.c:
		- only expose explicitly exported types to importer
		- fix behind your back
	go.h:
		- add deep() prototype (fixes 64-bit linux crash on time.go)
	go.y:
		- add a new syntax error case
	walk.c:
		- allow a,b = f() where f is func ptr (fixes bug088)

R=ken
OCL=15617
CL=15630
2008-09-22 12:45:01 -07:00
Rob Pike
9ebd9d3b7d fix some tests
R=ken
OCL=15598
CL=15598
2008-09-20 15:16:48 -07:00
Robert Griesemer
6a0fb60866 - added item to TODO list in go_spec
- filed a bug

R=r
OCL=15517
CL=15517
2008-09-18 23:16:22 -07:00
Rob Pike
85f8d4501a fix some broken tests (tests themselves were wrong)
R=gri,rsc
OCL=15310
CL=15402
2008-09-16 14:03:43 -07:00
Rob Pike
c3b397bd99 var bug is fixed
R=rsc
OCL=15400
CL=15400
2008-09-16 14:02:41 -07:00
Rob Pike
6f38063369 update bugs
delete bug100.go since the compiler is doing what we agreed.

R=gri
OCL=15367
CL=15367
2008-09-15 14:27:15 -07:00
Russ Cox
22731159b0 array literals not initialized
R=r
DELTA=11  (11 added, 0 deleted, 0 changed)
OCL=15222
CL=15276
2008-09-12 16:35:02 -07:00