Based on the observation that a great number of the types that
are copied or compared in interfaces, maps, and channels are
word-sized, this uses specialized copy and equality functions
for them that use a word instead of 4 or 8 bytes. Seems to yield
0-6% improvements in performance in the benchmarks I've run.
For example, with the regexp benchmarks:
Before:
regexp.BenchmarkLiteral 500000 3.26 µs/op
regexp.BenchmarkNotLiteral 100000 13.67 µs/op
regexp.BenchmarkMatchClass 100000 18.72 µs/op
regexp.BenchmarkMatchClass_InRange 100000 20.04 µs/op
regexp.BenchmarkReplaceAll 100000 27.85 µs/op
After:
regexp.BenchmarkLiteral 500000 3.11 µs/op
regexp.BenchmarkNotLiteral 200000 13.29 µs/op
regexp.BenchmarkMatchClass 100000 17.65 µs/op
regexp.BenchmarkMatchClass_InRange 100000 18.49 µs/op
regexp.BenchmarkReplaceAll 100000 26.34 µs/op
R=rsc
CC=golang-dev
https://golang.org/cl/1967047
Cannot use paren field in Node because all
instances of a given symbol name use the same Node.
Fixes#1022.
R=ken2
CC=golang-dev
https://golang.org/cl/2015043
x.go:13: cannot use t (type T) as type Reader in assignment:
T does not implement Reader (Read method requires pointer receiver)
x.go:19: cannot use q (type Q) as type Reader in assignment:
Q does not implement Reader (missing Read method)
have read()
want Read()
x.go:22: cannot use z (type int) as type Reader in assignment:
int does not implement Reader (missing Read method)
x.go:24: too many arguments to conversion to complex: complex(1, 3)
R=ken2
CC=golang-dev
https://golang.org/cl/1736041
* Code for assignment, conversions now mirrors spec.
* Changed some snprint -> smprint.
* Renamed runtime functions to separate
interface conversions from type assertions:
convT2I, assertI2T, etc.
* Correct checking of \U sequences.
Fixes#840.
Fixes#830.
Fixes#778.
R=ken2
CC=golang-dev
https://golang.org/cl/1303042
cannot allocate an audomatic temp
while real registers are allocated.
there is a chance that the automatic
will be allocated to one of the
allocated registers. the fix is to
not registerize such variables.
R=rsc
CC=golang-dev
https://golang.org/cl/1202042
main semantic change is to enforce single argument to panic.
runtime: change to 1-argument panic.
use String method on argument if it has one.
R=ken2, r
CC=golang-dev
https://golang.org/cl/812043
bonus: type switch now detects multiple uses of identical interface types.
bonus: interface types are now order-independent, following the spec.
R=ken2
CC=golang-dev
https://golang.org/cl/194053
5g/6g/8g: add import statements to export metadata, mapping package path to package name.
recognize "" as the path of the package in export metadata.
use "" as the path of the package in object symbol names.
5c/6c/8c, 5a/6a/8a: rewrite leading . to "". so that ·Sin means Sin in this package.
5l/6l/8l: rewrite "" in symbol names as object files are read.
gotest: handle new symbol names.
gopack: handle new import lines in export metadata.
Collectively, these changes eliminate the assumption of a global
name space in the object file formats. Higher level pieces such as
reflect and the computation of type hashes still depend on the
assumption; we're not done yet.
R=ken2, r, ken3
CC=golang-dev
https://golang.org/cl/186263
* switch to real dot (.) instead of center dot (·) everywhere in object files.
before it was half and half depending on where in the name it appeared.
* in 6c/6a/etc identifiers, · can still be used but turns into . immediately.
* in export metadata, replace package identifiers with quoted strings
(still package names, not paths).
R=ken2, r
CC=golang-dev
https://golang.org/cl/190076
because they are in package runtime.
another step to enforcing package boundaries.
R=r
DELTA=732 (114 added, 93 deleted, 525 changed)
OCL=35811
CL=35824
references during the parsing of :=. the base
problem is that when reading
a,b,c,d
the parser makes those refer to existing variables,
which might create a few stub top-level ones
for undefined names, but then if a := is the next
token, we need to undo those stubs.
this was causing problems in multifile packages
in which one file used a := variable named rpc
and the other imported a package named rpc.
R=ken
OCL=35446
CL=35446
introduce NodeList* type in compiler to replace OLIST.
this clarifies where lists can and cannot occur.
list append and concatenation are now cheap.
the _r rules are gone from yacc.
rev and unrev are gone.
no more lists of lists.
the representation of assignments is a bit clunkier.
split into OAS (1=1) and OAS2 (2 or more on one side).
delete dead chanrecv3 code.
delay construction of func types.
R=ken
OCL=31745
CL=31762
walkstate -> walkstmt
walktype -> walkexpr; stmts moved to walkstmt
walktype and friends have a final Node **init
argument that can have side effects appended,
making it more explicit when they do and do not happen.
this replaces the old global addtop and addtotop.
delete switch map and interface conversion cases
(dropped from the language months ago).
R=ken
OCL=31465
CL=31468
* give genwrapper and genembedtramp the same signature.
* move duint8, duint16, duint32, duint64, duintptr into gc.
* tidy genwrapper.
* bug involving struct field symbols in signature list.
(hash-order dependent so hard to trigger)
* new Type print format %#-T like %#T but omits
names on function arguments.
R=ken
OCL=31237
CL=31237
do not bother warning about marks left
on stack after syntax errors.
leave OCONV nodes in tree to avoid type errors
arising from multiple walks.
R=ken
OCL=30639
CL=30662
* 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
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
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
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
store only the original import path string (+ .a)
if 6g resolves it to an archive file.
let 6l re-resolve the .a at link time.
this lets libraries build against an archive
in the current directory but get used
against an installed archive.
R=r
OCL=27244
CL=27244