likely to go back to registers for most temporaries.
most tests in lib pass. these fail:
datafmt
fmt
go/scanner
log
reflect
strconv
template
R=ken
OCL=29896
CL=29898
* floating point -> integer conversions.
x86 defines that overflow/underflow
results in 1<<15, 1<<31, 1<<63 for
int16, int32, int64. when building the
unsigned conversions out of the native signed
ones, 8g turns overflow/underflow into zero.
the spec does not say what should happen.
* many tiny bug fixes. can run a large number
of files from go/test now, and can fmt.Printf.
* struggling with byte register allocation
and float32 computation.
R=ken
OCL=29642
CL=29811
* conversions all in one place.
* no separate load, store phases;
direct memory addressing when possible
(this is the x86 after all!).
avoids extra registers, extra MOVQs.
* fixes int32 -> uint64 bug
(was zero-extending)
R=ken
OCL=29482
CL=29484
* string format changed
* files got renamed
* new files that i forgot to check in last time
updates are all copy and paste from 6g
R=ken
OCL=29385
CL=29400
structtype{a:1, a:2}
maptypetype{"xx":1, "xx":2}
arraytypetype{5:1, 5:2}
2. bug in registerization concerning
alias of a struct and one of its elements
3. code optimization of struct.field
(which exposed bug in 2)
R=r
OCL=29315
CL=29315
*** Reason for rollback ***
too many files included
*** Original change description ***
simplifying grammar: delete LBASETYPE and LACONST
R=ken
OCL=29303
CL=29303
when assigning a multifield object
(structs or arrays of structs) they
must not contain any fields that could
not be assigned individually.
R=ken
OCL=29192
CL=29194
this is not a user-visible change.
before, all interface values were
struct Itype {
Sigt *type;
Sigi *inter;
void *method[n];
}
struct Iface {
void *addr;
Itype *itype;
}
the itype is basically a vtable, but it's unnecessary
if the static type is interface{ }.
for interface values with static type empty, the
new representation is
struct Eface {
void *addr;
Sigt *type;
}
this complicates the code somewhat, but
it reduces the number of Itypes that
have to be computed and cached,
it opens up opportunities to avoid function
calls in a few common cases,
and it will make it possible to lay out
interface{} values at compile time,
which i think i'll need for the new reflection.
R=ken
OCL=28701
CL=29121
do not update lineno from ONAME nodes,
because they have declaration lineno not use.
show actual name in top-level statement error.
before
runtime.a:7: x.go:5: walkstate: NAME not a top level statement
after
x.go:14: walkstate: runtime.Goexit not a top level statement
R=ken
OCL=29113
CL=29116
if both types are named, they must be
the same type (arising from the same
declaration).
R=r,gri
DELTA=44 (21 added, 4 deleted, 19 changed)
OCL=28436
CL=28577
with method m, s.m() is ok and m now shows up
in s's method set for interface runtime.
see http://cl/28419-p2 for new test interface10.go.
R=ken
OCL=28420
CL=28423
replace "shape error across CALL" with more information.
x.go:7: not enough arguments to CALL
a int, b int
int
x.go:10: assignment count mismatch: 3 = 2
x.go:12: too many arguments to RETURN
[no arguments expected]
int, int, int
also leave type alone after conversion failure,
for later errors:
bug049.go:6: cannot convert nil constant to string
bug049.go:6: illegal types for operand: EQ
string
nil # this used to be blank
R=ken
OCL=28405
CL=28407
TEXT tramp
MOVQ 8(SP), AX
ADDQ $40, AX
MOVQ AX, 8(SP)
JMP oldfunc
is now
TEXT tramp
ADDQ $40, 8(SP)
JMP oldfunc
and if s/40/0/, then it simplifies to
TEXT tramp
JMP oldfunc
(the tramp is still needed to satisfy
symbol references from other object files)
R=ken
OCL=28377
CL=28381
so that 6l can discard strings used by dead code.
also, for short strings, generate DUPOK global
symbols so that references to, say, "%s: %s" in
multiple files result in only a single symbol.
R=ken
OCL=28361
CL=28361
* byteastring is no longer used
* do not generate ODCL, OAS for globals
(wasn't generating any code but might
save one or two init functions)
* do not call self from Init function
R=ken
OCL=28309
CL=28309
editing the firstp list was ineffective,
because follow rebuilds it from the textp list.
the symbols for dead code were being dropped
from the binary but the code was all still there.
text for fmt.Printf("hello, world\n") drops
from 143945 to 128650.
R=r,ken
DELTA=22 (20 added, 0 deleted, 2 changed)
OCL=28255
CL=28290
of 5g. 5l balks at the output and running 5g with -S shows
the true extent of the disaster. Still, better than
yesterday. Maybe.
Tested on the canonical:
package main
func main() {
}
R=rsc
APPROVED=rsc
DELTA=4182 (4181 added, 0 deleted, 1 changed)
OCL=27601
CL=28175
exact spec:
a) must be a multi-assignment w :=
b) a proper subset of the lhs
can be declared in same block
with the same type with no
"redeclaration" error
R=r
OCL=27610
CL=27610
this allows gotest to find the locally built package when doing
make
gotest
without this option, one would have to say
make install
gotest
which kinda defeats the purpose
based on discussions with rsc.
R=ken,rsc
DELTA=12 (10 added, 1 deleted, 1 changed)
OCL=27606
CL=27606
