1. dont clear external and heap objects
2. propagate constant assignment

R=r
OCL=28254
CL=28254

It now exports a Var interface (anyone can export their own custom var types now), so users need to create and manage their own vars and mark them as exportable via the Publish function. They are exposed via /debug/vars.

R=r,rsc
APPROVED=r
DELTA=605  (314 added, 186 deleted, 105 changed)
OCL=28143
CL=28239

  (this corresponds to the currently running version of godoc)

R=r
DELTA=11  (0 added, 10 deleted, 1 changed)
OCL=28238
CL=28238

R=r
OCL=28204
CL=28204

R=r
OCL=28201
CL=28201

R=gri
APPROVED=gri
DELTA=1  (0 added, 0 deleted, 1 changed)
OCL=28184
CL=28192

value has been received.  This failed with gccgo before
today.

R=ken,r
DELTA=18  (18 added, 0 deleted, 0 changed)
OCL=28185
CL=28190

- template-driven ast printing now can successfully
  reproduce entire Go programs

next steps:
- fine-tuning of output
- print interspersed comments
- cleanup and testing against all Go programs
- replace astprinter

R=r
OCL=28181
CL=28181

drip init function if it doesnt do anything

R=r
OCL=28180
CL=28180

better calling of morestack

R=r
OCL=28179
CL=28179

R=r
DELTA=16  (0 added, 3 deleted, 13 changed)
OCL=28133
CL=28178

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

- fixed a bug and added corresponding test case

R=rsc
DELTA=114  (18 added, 29 deleted, 67 changed)
OCL=28114
CL=28128

R=ken
OCL=28120
CL=28124

R=r
DELTA=12  (9 added, 0 deleted, 3 changed)
OCL=28096
CL=28118

R=r
DELTA=30  (9 added, 1 deleted, 20 changed)
OCL=28104
CL=28117

to match traditional c linkers.

R=r
DELTA=42  (8 added, 12 deleted, 22 changed)
OCL=28101
CL=28115

TBR=rsc
DELTA=2  (1 added, 0 deleted, 1 changed)
OCL=28066
CL=28070

add test for structure alignment/offset.

R=gri
DELTA=49  (35 added, 0 deleted, 14 changed)
OCL=28068
CL=28068

you should be able to convert a pointer to an array to a
slice, you should not be able to convert an array to a slice.
Currently 6g works the other way around.

R=ken,rsc
DELTA=17  (17 added, 0 deleted, 0 changed)
OCL=28033
CL=28067

R=r
DELTA=4  (3 added, 1 deleted, 0 changed)
OCL=28063
CL=28065

don't relay the signal when restarting it.

R=r
DELTA=1  (0 added, 0 deleted, 1 changed)
OCL=28060
CL=28064

iant: will this be ok in gccgo?

R=rsc
DELTA=9  (8 added, 0 deleted, 1 changed)
OCL=28059
CL=28062

R=rsc
DELTA=1  (0 added, 0 deleted, 1 changed)
OCL=28056
CL=28058

one fd gets too many EPIPEs in a row
during write.

R=r
DELTA=10  (9 added, 0 deleted, 1 changed)
OCL=28057
CL=28057

Currently a http server will be killed on receipt of SIGPIPE
if a client closes a socket which the server is trying to
write to.

R=rsc
APPROVED=rsc
DELTA=2  (0 added, 0 deleted, 2 changed)
OCL=27959
CL=28055

which hit EAGAIN).

R=rsc
APPROVED=rsc
DELTA=2  (0 added, 0 deleted, 2 changed)
OCL=27955
CL=28054

R=tgs
DELTA=1  (0 added, 0 deleted, 1 changed)
OCL=28032
CL=28032

- work-around for incorrect import path
- added tmpl root in order to run against a goroot w/o templates
- clarifications

Daily snapshot of syntax-driven formatter. Some progress.

Updated gccgo Makefile.

TBR=r
OCL=28004
CL=28004

bug found by anton

R=r
OCL=28001
CL=28001

R=r
OCL=27987
CL=27987

R=r
OCL=27980
CL=27980

R=r
OCL=27979
CL=27979

TBR=r
DELTA=1  (0 added, 0 deleted, 1 changed)
OCL=27969
CL=27969

declbad.go:15:3: error: variables redeclared but no variable is new
declbad.go:20:3: error: redefinition of 'f'
declbad.go:19:3: note: previous definition of 'f' was here
declbad.go:25:3: error: redefinition of 'i'
declbad.go:24:3: note: previous definition of 'i' was here
declbad.go:30:3: error: variables redeclared but no variable is new
declbad.go:35:3: error: redefinition of 'i'
declbad.go:34:3: note: previous definition of 'i' was here
declbad.go:40:3: error: variables redeclared but no variable is new
declbad.go:45:3: error: variables redeclared but no variable is new

R=r
DELTA=10  (0 added, 0 deleted, 10 changed)
OCL=27934
CL=27957

R=r
APPROVED=r
DELTA=20  (11 added, 6 deleted, 3 changed)
OCL=27782
CL=27950

This makes "make nuke install" work again.

R=r
APPROVED=r
DELTA=2  (1 added, 0 deleted, 1 changed)
OCL=27929
CL=27932

R=rsc
OCL=27928
CL=27928

R=rsc
APPROVED=r
DELTA=11  (10 added, 0 deleted, 1 changed)
OCL=27723
CL=27879

R=r
APPROVED=r
DELTA=62  (58 added, 1 deleted, 3 changed)
OCL=27756
CL=27877