Also improve error message for unsupported integer types

R=agl, rsc
CC=golang-dev
https://golang.org/cl/4873049

forminfo.go:
- Wrappers for table data.
- Per Form dispatch table.
composition.go:
- reorderBuffer type.  Implements decomposition, reordering, and composition.
- Note: decompose and decomposeString fields in formInfo could be replaced by
  a pointer to the trie for the respective form.  The proposed design makes
  testing easier, though.
normalization.go:
- Temporarily added panic("not implemented") methods to make the tests run.
  These will be removed again with the next CL, which will introduce the
  implementation.

R=r, rogpeppe, mpvl, rsc
CC=golang-dev
https://golang.org/cl/4875043

R=golang-dev, dsymonds
CC=golang-dev
https://golang.org/cl/4911041

Replaces the toy func Reverse(*Template) with one that implements
naive autoescaping.

Now Escape(*Template) walks a template parse tree to find all
template actions and adds the |html command to them if it is not
already present.

R=golang-dev, r
CC=golang-dev
https://golang.org/cl/4867049

R=gri, r, rsc
CC=golang-dev
https://golang.org/cl/4891041

(Leave exp/template/html where it is for now.)

R=golang-dev, dsymonds
CC=golang-dev
https://golang.org/cl/4899048

It's already in old/template; make that build.
Update a couple of references to point to the old template.
They can be updated later.
Update goplay to use exp/template.

R=golang-dev, dsymonds
CC=golang-dev
https://golang.org/cl/4902046

R=golang-dev, dsymonds
CC=golang-dev
https://golang.org/cl/4876057

This is just moving the URL code from package http into its own package,
which has been planned for a while.
Besides clarity, this also breaks a nascent dependency cycle the new template
package was about to introduce.

Add a gofix module, url, and use it to generate changes outside http and url.

Sadness about the churn, gladness about some of the naming improvements.

R=dsymonds, bradfitz, rsc, gustavo, r
CC=golang-dev
https://golang.org/cl/4893043

Fixes #2148.

R=golang-dev, bradfitz, rsc
CC=golang-dev
https://golang.org/cl/4905042

Fixes #2156.

R=rsc
CC=golang-dev
https://golang.org/cl/4899046

R=golang-dev, gri
CC=golang-dev
https://golang.org/cl/4901047

Add multicast stubs and sync with recent changes.

R=golang-dev, mikioh.mikioh
CC=golang-dev
https://golang.org/cl/4838058

Multicast address handling is not consistent across all BSDs. Move
the multicast address handling code into OS dependent files. This
will be needed for OpenBSD support.

R=mikioh.mikioh, golang-dev
CC=golang-dev
https://golang.org/cl/4809074

Fixes #2127.

R=rsc
CC=golang-dev
https://golang.org/cl/4873043

The corruption can occur when GOMAXPROCS
is changed from >1 to 1, since GOMAXPROCS=1
does not imply there is only 1 goroutine running,
other goroutines can still be not parked after
the change.

R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/4873050

When the C API being used includes multiple names for the same
underlying symbol (e.g. multiple #define's for the same variable), then
cgo will generate the same placeholder variables for each name.  This
then prevents the code from compiling due to multiple declarations of
the same variable - so change cgo to only create one instance of the
variable for the underlying symbol.

R=rsc
CC=golang-dev
https://golang.org/cl/4826055

R=golang-dev
CC=golang-dev, rsc
https://golang.org/cl/4837044

src/cmd/5c/reg.c:
. Added USED() attribute.

src/cmd/6c/cgen.c:
. Revised code around "REGARG" to resemble use in "8c" and
  consequently remove a warning.

src/cmd/6l/asm.c:
. Added USED() attributes.
. Removed an unnecessary assignment.

R=golang-dev
CC=golang-dev, rsc
https://golang.org/cl/4836045

R=golang-dev
CC=golang-dev, rsc
https://golang.org/cl/4835043

// ERROR "pattern1" "pattern2"

means that there has to be one or more
lines matching pattern1 and then excluding
those, there have to be one or more lines
matching pattern2.  So if you expect two
different error messages from a particular
line, writing two separate patterns checks
that both errors are produced.

Also, errchk now flags lines that produce
more errors than expected.  Before, as long as
at least one error matched the pattern, all the
others were ignored.

Revise tests to expect or silence these
additional errors.

R=lvd, r, iant
CC=golang-dev
https://golang.org/cl/4869044

Currently it's possible to write:
var s rpc.Server
...
// reuse for my own purposes
s.Lock()
...
s.Unlock()
which is seemingly not intended.

R=golang-dev, adg
CC=golang-dev
https://golang.org/cl/4888049

They are seemingly not intended to be a part
of the public interface.

R=golang-dev, adg
CC=golang-dev
https://golang.org/cl/4873052

R=golang-dev, dvyukov
CC=golang-dev
https://golang.org/cl/4896046

R=golang-dev, adg, dsymonds
CC=golang-dev
https://golang.org/cl/4875050

R=r, adg
CC=golang-dev
https://golang.org/cl/4904045

- show documentation in text mode; e.g.:
http://localhost:7777/cmd/godoc/?f=text

- simplify template by using $ variable

R=r, adg
CC=golang-dev
https://golang.org/cl/4873045

- HTML-escape URL paths
- URL-escape URL parameters

R=bradfitz
CC=golang-dev
https://golang.org/cl/4890041

This is useful for applications that want to micromanage the rpc service.
Moved part of ServeCodec into a new readRequest function.
Renamed existing readRequest to readRequestHeader, and reordered
its parameters to align with the new readRequest and service.call.

R=golang-dev, r, rsc, sougou
CC=golang-dev, msolomon
https://golang.org/cl/4889043

Do this by adding a fake package called builtin. At the moment, godoc will
not present this documentation because the function names are all lower case,
but there are plans to address this.
The print and println functions are undocumented here. I could be talked
into doing them but I'd rather not promote their use.

R=dsymonds, n13m3y3r, r, gri, rsc
CC=golang-dev
https://golang.org/cl/4907041

R=rsc
CC=golang-dev
https://golang.org/cl/4904044

CL 4873048 introduced the ability to build without hg and
getting an "unknown" version.  While this approach works
to avoid the hg dependency, it also means that every
exported tree that is built without hg or .hg will have not
only missing information, but will also be compatible to
one another.  Considering that it is a common practice to
remove the VCS data in distributions, I suggest we don't
take this approach to avoid its consequences.

This CL fixes the same problem in a different way: if a
VERSION file at the top of the tree exists, use it at
all times.  If it doesn't, fall back to using information
from hg necessarily, and fail if that's not possible.  The
error message when VERSION and hg are not available
instructs users to handle it properly.

The VERSION file can be generated with
"src/version.bash -save" while hg is still around.

R=golang-dev, rsc, gustavo
CC=golang-dev
https://golang.org/cl/4897043

This was initially pushed as part of CL 4876046, found
when logic in exp/template was using the method on
an Invalid value.

R=rsc
CC=golang-dev
https://golang.org/cl/4890043

R=golang-dev, dsymonds, n13m3y3r
CC=golang-dev
https://golang.org/cl/4873048

Fixes #2152.

R=golang-dev, dvyukov
CC=golang-dev
https://golang.org/cl/4888044

First step of moving exp/template into template: save the old code.
Code is unedited except for target name in Makefile.

R=golang-dev, dsymonds
CC=golang-dev
https://golang.org/cl/4893042

While using exp/template in practice, the syntax for
indexing values using the "index" action was found to be
very inconvenient for frequent use when handling dynamic
data from maps such as the ones used with json and yaml,
that use a type like map[string]interface{}.

For these kinds of maps, the default handling of fields as
{{.Field}} makes the task of handling the several references
significantly more pleasant and elegant, and is equivalent
to what's currently done in the "template" package and in
other external packages (e.g. mustache).

Even with this change, the index action is still relevant
as it allows indexing maps in other scenarios where keys
wouldn't be valid field names.

R=golang-dev, r, gustavo
CC=golang-dev
https://golang.org/cl/4898043

This avoids a non-obvious panic when range is used on a
nil interface, and fixes it by behaving as if the range
was empty.

The new behavior is equivalent to the outcome of iterating
on a nil map or slice, and is useful because it allows
generic structures such as used in json (map[string]interface{})
to behave correctly if a key generally set to a list or map
isn't present.

R=golang-dev, r, gustavo
CC=golang-dev
https://golang.org/cl/4876046

R=golang-dev, dsymonds
CC=golang-dev
https://golang.org/cl/4895042

R=golang-dev, dsymonds
CC=golang-dev
https://golang.org/cl/4875048