This is a semantic but no API change. It is a cleaner
implementation of pure filtering. Applications that
need function bodies stripped can easily do this them-
selves.
R=rsc
CC=golang-dev
https://golang.org/cl/5206046
This CL introduces the go.Example type and go.Examples functions that
are used to represent and extract code samples from Go source.
They should be of the form:
// Output of this function.
func ExampleFoo() {
fmt.Println("Output of this function.")
}
It also modifies godoc to read example code from _test.go files,
and include them in the HTML output with JavaScript-driven toggles.
It also implements testing of example functions with gotest.
The stdout/stderr is compared against the output comment on the
function.
This CL includes examples for the sort.Ints function and the
sort.SortInts type. After patching this CL in and re-building go/doc
and godoc, try
godoc -http=localhost:6060
and visit http://localhost:6060/pkg/sort/
R=gri, r, rsc
CC=golang-dev
https://golang.org/cl/5137041
The malloc sample trigger was not being set in a
new m, so the first allocation in each new m - the
goroutine structure - was being sampled with
probability 1 instead of probability sizeof(G)/rate,
an oversampling of about 5000x for the default
rate of 1 MB. This bug made pprof graphs show
far more G allocations than there actually were.
R=golang-dev, r
CC=golang-dev
https://golang.org/cl/5224041
Fixes#2337.
Unfortunate sequence of events is:
1. maxcpu=2, mcpu=1, grunning=1
2. starttheworld creates an extra M:
maxcpu=2, mcpu=2, grunning=1
4. the goroutine calls runtime.GOMAXPROCS(1)
maxcpu=1, mcpu=2, grunning=1
5. since it sees mcpu>maxcpu, it calls gosched()
6. schedule() deschedules the goroutine:
maxcpu=1, mcpu=1, grunning=0
7. schedule() call getnextandunlock() which
fails to pick up the goroutine again,
because canaddcpu() fails, because mcpu==maxcpu
8. then it sees that grunning==0,
reports deadlock and terminates
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/5191044
therefore unlikely that there is a good use for its string version
LastBoundaryInString. Yet, the implemenation of this method would complicate
things a bit as it would require the introduction for another interface and
some duplication of code. Removing it seems a better choice.
R=r
CC=golang-dev
https://golang.org/cl/5182044
Major changes between hybi-08 and hybi-13
- hybi-08 uses Sec-WebSocket-Origin, but hybi-13 uses Origin
- hybi-13 introduces new close status codes.
hybi-17 spec (editorial changes of hybi-13) mentions
- if a server doesn't support the requested version, it MUST respond
with Sec-WebSocket-Version headers containing all available versions.
- client MUST close the connection upon receiving a masked frame
- server MUST close the connection upon receiving a non-masked frame
note that hybi-17 still uses "Sec-WebSocket-Version: 13"
see http://code.google.com/p/pywebsocket/wiki/WebSocketProtocolSpec
for changes between spec drafts.
R=golang-dev, adg
CC=golang-dev
https://golang.org/cl/5147043
This can work only if there is no type info required to initialize the decoder,
but it's easy and gains a few percent in the basic benchmarks by avoiding
bufio when it's a bytes.Buffer - a testing-only scenario, I admit.
Add a comment about what Decode expects from the input.
R=rsc
CC=golang-dev
https://golang.org/cl/5165048
When ncpu < 2, work.nproc is always 1 which results in infinite helper
threads being created if gomaxprocs > 1 and MaxGcproc > 1. Avoid this
by using the same limits as imposed helpgc().
R=golang-dev, rsc, dvyukov
CC=golang-dev
https://golang.org/cl/5176044
This change adds the osyield and usleep
functions and code to read the number of
processors from /dev/sysstat.
I also changed SysAlloc to return nil
when brk fails (it was returning -1).
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/5177049
The spin-off renames some types. The new names are simply better:
image.Color -> color.Color
image.ColorModel -> color.Model
image.ColorModelFunc -> color.ModelFunc
image.PalettedColorModel -> color.Palette
image.RGBAColor -> color.RGBA
image.RGBAColorModel -> color.RGBAModel
image.RGBA64Color -> color.RGBA64
image.RGBA64ColorModel -> color.RGBA64Model
(similarly for NRGBAColor, GrayColorModel, etc)
The image.ColorImage type stays in the image package, but is renamed:
image.ColorImage -> image.Uniform
The image.Image implementations (image.RGBA, image.RGBA64, image.NRGBA,
image.Alpha, etc) do not change their name, and gain a nice symmetry:
an image.RGBA is an image of color.RGBA, etc.
The image.Black, image.Opaque uniform images remain unchanged (although
their type is renamed from image.ColorImage to image.Uniform). The
corresponding color types (color.Black, color.Opaque, etc) are new.
Nothing in the image/ycbcr is renamed yet. The ycbcr.YCbCrColor and
ycbcr.YCbCrImage types will eventually migrate to color.YCbCr and
image.YCbCr, but that will be a separate CL.
R=r, bsiegert
CC=golang-dev
https://golang.org/cl/5132048
This implements a replacer for when all old strings are single
bytes, but new values are not.
BenchmarkHTMLEscapeNew 1000000 1090 ns/op
BenchmarkHTMLEscapeOld 1000000 2049 ns/op
R=rsc
CC=golang-dev
https://golang.org/cl/5176043
When all old & new string values are single bytes,
byteReplacer is now used, instead of the generic
algorithm.
BenchmarkGenericMatch 10000 102519 ns/op
BenchmarkByteByteMatch 1000000 2178 ns/op
fast path, when nothing matches:
BenchmarkByteByteNoMatch 1000000 1109 ns/op
comparisons to multiple Replace calls:
BenchmarkByteByteReplaces 100000 16164 ns/op
comparison to strings.Map:
BenchmarkByteByteMap 500000 5454 ns/op
R=rsc
CC=golang-dev
https://golang.org/cl/5175050
The map implementation was using the C idiom of using
a pointer just past the end of its table as a limit pointer.
Unfortunately, the garbage collector sees that pointer as
pointing at the block adjacent to the map table, pinning
in memory a block that would otherwise be freed.
Fix by making limit pointer point at last valid entry, not
just past it.
Reviewed by Mike Burrows.
R=golang-dev, bradfitz, lvd, r
CC=golang-dev
https://golang.org/cl/5158045
Running test/garbage/parser.out.
On a 4-core Lenovo X201s (Linux):
31.12u 0.60s 31.74r 1 cpu, no atomics
32.27u 0.58s 32.86r 1 cpu, atomic instructions
33.04u 0.83s 27.47r 2 cpu
On a 16-core Xeon (Linux):
33.08u 0.65s 33.80r 1 cpu, no atomics
34.87u 1.12s 29.60r 2 cpu
36.00u 1.87s 28.43r 3 cpu
36.46u 2.34s 27.10r 4 cpu
38.28u 3.85s 26.92r 5 cpu
37.72u 5.25s 26.73r 6 cpu
39.63u 7.11s 26.95r 7 cpu
39.67u 8.10s 26.68r 8 cpu
On a 2-core MacBook Pro Core 2 Duo 2.26 (circa 2009, MacBookPro5,5):
39.43u 1.45s 41.27r 1 cpu, no atomics
43.98u 2.95s 38.69r 2 cpu
On a 2-core Mac Mini Core 2 Duo 1.83 (circa 2008; Macmini2,1):
48.81u 2.12s 51.76r 1 cpu, no atomics
57.15u 4.72s 51.54r 2 cpu
The handoff algorithm is really only good for two cores.
Beyond that we will need to so something more sophisticated,
like have each core hand off to the next one, around a circle.
Even so, the code is a good checkpoint; for now we'll limit the
number of gc procs to at most 2.
R=dvyukov
CC=golang-dev
https://golang.org/cl/4641082
This is a possible optimization. I'm not sure the complexity is worth it.
The new benchmark in escape_test is 46us without and 35us with the optimization.
R=nigeltao
CC=golang-dev
https://golang.org/cl/5168041
This removes a few cases from escapeAction and clarifies the
responsibilities of urlFilter which no longer does any
escaping or normalization. It is now solely a filter.
R=nigeltao
CC=golang-dev
https://golang.org/cl/5162043
The letter is a holdover from C and unnecessary in Go.
Gofix module included.
Fixes#2306.
R=golang-dev, gri, dsymonds
CC=golang-dev
https://golang.org/cl/5158043
HTML5 allows embedded SVG and MathML.
Code searches show SVG is used for graphing.
This changes transition to deal with constructs like
<svg xmlns:xlink="http://www.w3.org/1999/xlink">
It changes attr and clients to call a single function that combines
the name lookup and "on" prefix check to determine an attribute
value type given an attribute name.
That function uses heuristics to recognize that
xlink:href and svg:href
have URL content, and that data-url is likely contains URL content,
since "javascript:" injection is such a problem.
I did a code search over a closure templates codebase to determine
patterns of custom attribute usage. I did something like
$ find . -name \*.soy | \
xargs egrep perl -ne 'while (s/\b((data-|\w+:)\w+)\s*=//) { print "$1\n"; }' | \
sort | uniq
to produce the list at the bottom.
Filtering that by egrep -i 'src|url|uri' produces
data-docConsumptionUri
data-docIconUrl
data-launchUrl
data-lazySrc
data-pageUrl
data-shareurl
data-suggestServerUrl
data-tweetUrl
g:secondaryurls
g:url
which seem to match all the ones that are likely URL content.
There are some short words that match that heuristic, but I still think it decent since
any custom attribute that has a numeric or enumerated keyword value will be unaffected by
the URL assumption.
Counterexamples from /usr/share/dict:
during, hourly, maturity, nourish, purloin, security, surly
Custom attributes present in existing closure templates codebase:
buzz:aid
data-a
data-action
data-actor
data-allowEqualityOps
data-analyticsId
data-bid
data-c
data-cartId
data-categoryId
data-cid
data-command
data-count
data-country
data-creativeId
data-cssToken
data-dest
data-docAttribution
data-docConsumptionUri
data-docCurrencyCode
data-docIconUrl
data-docId
data-docPrice
data-docPriceMicros
data-docTitle
data-docType
data-docid
data-email
data-entityid
data-errorindex
data-f
data-feature
data-fgid
data-filter
data-fireEvent
data-followable
data-followed
data-hashChange
data-height
data-hover
data-href
data-id
data-index
data-invitable
data-isFree
data-isPurchased
data-jid
data-jumpid
data-launchUrl
data-lazySrc
data-listType
data-maxVisiblePages
data-name
data-nid
data-nodeid
data-numItems
data-numPerPage
data-offerType
data-oid
data-opUsesEquality
data-overflowclass
data-packageName
data-pageId
data-pageUrl
data-pos
data-priceBrief
data-profileIds
data-query
data-rating
data-ref
data-rentalGrantPeriodDays
data-rentalactivePeriodHours
data-reviewId
data-role
data-score
data-shareurl
data-showGeLe
data-showLineInclude
data-size
data-sortval
data-suggestServerType
data-suggestServerUrl
data-suggestionIndex
data-tabBarId
data-tabBarIndex
data-tags
data-target
data-textColor
data-theme
data-title
data-toggletarget
data-tooltip
data-trailerId
data-transactionId
data-transition
data-ts
data-tweetContent
data-tweetUrl
data-type
data-useAjax
data-value
data-width
data-x
dm:index
dm:type
g:aspects
g:decorateusingsecondary
g:em
g:entity
g:groups
g:id
g:istoplevel
g:li
g:numresults
g:oid
g:parentId
g:pl
g:pt
g:rating_override
g:secondaryurls
g:sortby
g:startindex
g:target
g:type
g:url
g:value
ga:barsize
ga:css
ga:expandAfterCharsExceed
ga:initialNumRows
ga:nocancelicon
ga:numRowsToExpandTo
ga:type
ga:unlockwhenrated
gw:address
gw:businessname
gw:comment
gw:phone
gw:source
ng:controller
xlink:href
xml:lang
xmlns:atom
xmlns:dc
xmlns:jstd
xmlns:ng
xmlns:og
xmlns:webstore
xmlns:xlink
R=nigeltao
CC=golang-dev
https://golang.org/cl/5119041
The normalization that prevents element name and comment injection in
<{{.}}
by converting it to
<{{.}}
breaks
<!DOCTYPE html>
Instead of splitting states to have a start of document state and a text
state, I whitelist <!DOCTYPE.
R=nigeltao
CC=golang-dev
https://golang.org/cl/5131051
This is just a new API to do many replacements at once.
While the point of this API is to be faster than doing replacements one
at a time, the implementation in this CL has the optimizations removed
and may actually be slower.
Future CLs will bring back & add optimizations.
R=r, rsc, rogpeppe
CC=golang-dev
https://golang.org/cl/5081042
In
{{$x := . | foo}}
{{$x}}
the first action is a variable assignment that contributes
nothing to the output while the first is a use that needs
to be escaped.
This CL fixes escapeAction to distinguish assignments from
interpolations and to only modify interpolations.
R=nigeltao, r
CC=golang-dev
https://golang.org/cl/5143048
CL 5040041 (https://golang.org/cl/5040041)
changed the use of []int to []int32 internally so
that encoding/binary could be used. This is no
longer needed (gobs can encode ints), and using
[]int is more in sync w/ the semantics of the data
structure (the index elements are indices which are
ints). Changing it back.
R=r
CC=golang-dev
https://golang.org/cl/5141049
Does some TODOs and changes the term "div" in an error message
to "division" to avoid confusion with "<div>".
R=nigeltao, r
CC=golang-dev
https://golang.org/cl/5141047
Case-insensitive strcmp without using ToLower.
(Using ToLower is not always correct, and it allocates.)
R=golang-dev, r
CC=golang-dev
https://golang.org/cl/5143044
This makes sure that all JS newlines are encoded in JSON.
It also moots a TODO about possibly escaping supplemental codepoints.
I served:
Content-Type: text/javascript;charset=UTF-8
var s = "%s";
document.write("<p>", s, "</p><ol>");
for (var i = 0; i < s.length; i++) {
document.write("<li>", s.charCodeAt(i).toString(16), "</li>");
}
document.write("</l>");
where %s was replaced with bytes "\xf0\x9d\x84\x9e" to test
straight UTF-8 instead of encoding surrogates separately.
Recent Firefox, Chrome, and Safari all decoded it properly.
I have yet to try it on IE or older versions.
R=nigeltao
CC=golang-dev
https://golang.org/cl/5129042
The template
"<a="
caused an infinite loop in escape text.
The change to tTag fixes that and the change to escape.go causes
escapeText to panic on any infinite loop that does not involve
a state cycle.
R=nigeltao
CC=golang-dev
https://golang.org/cl/5115041
HTML parsers may differ on whether
<input id= onchange=f( ends in id's or onchange's value,
<a class=`foo ends inside a value,
<input style=font:'Arial' needs open-quote fixup.
Per
http://www.w3.org/TR/html5/tokenization.html#attribute-value-unquoted-state
this treats the error cases in 8.2.4.40 Attribute value (unquoted) state
as fatal errors.
\> U+0022 QUOTATION MARK (")
\> U+0027 APOSTROPHE (')
\> U+003C LESS-THAN SIGN (<)
\> U+003D EQUALS SIGN (=)
\> U+0060 GRAVE ACCENT (`)
Parse error. Treat it as per the "anything else" entry below.
and emits ErrBadHTML.
R=nigeltao
CC=golang-dev
https://golang.org/cl/5085050
The Dwarf info has the full typenames, the go *struct runtime.commonType
has the short name. A more permanent fix would link the two together
but this way the user gets useable stack traces for now.
R=rsc
CC=golang-dev
https://golang.org/cl/5097046
One benefit of websocket is that it is full-duplex so that it could
send and receive at the same time.
This CL makes websocket goroutine safe, so user could use websocket
both on goroutine for read and on goroutine for write.
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/5058043
When templates are stored in external files, developers often embed
comments to explain&|disable code.
<!-- Oblique reference to project code name here -->
{{if .C}}...{{else}}<!-- commented out default -->{{end}}
This unnecessarily increases the size of shipped HTML and can leak
information.
This change elides all comments of the following types:
1. <!-- ... --> comments found in source.
2. /*...*/ and // comments found in <script> elements.
3. /*...*/ and // comments found in <style> elements.
It does not elide /*...*/ or // comments found in HTML attributes:
4. <button onclick="/*...*/">
5. <div style="/*...*/">
I can find no examples of comments in attributes in Closure Templates
code and doing so would require keeping track of character positions
post decode in
<button onclick="/*...*/">
To prevent token joining, /*comments*/ are JS and CSS comments are
replaced with a whitespace char.
HTML comments are not, but to prevent token joining we could try to
detect cases like
<<!---->b>
</<!---->b>
which has a well defined meaning in HTML but will cause a validator
to barf. This is difficult, and this is a very minor case.
I have punted for now, but if we need to address this case, the best
way would be to normalize '<' in stateText to '<' consistently.
The whitespace to replace a JS /*comment*/ with depends on whether
there is an embedded line terminator since
break/*
*/foo
...
is equivalent to
break;
foo
...
while
break/**/foo
...
is equivalent to
break foo;
...
Comment eliding can interfere with IE conditional comments.
http://en.wikipedia.org/wiki/Conditional_comment
<!--[if IE 6]>
<p>You are using Internet Explorer 6.</p>
<![endif]-->
/*@cc_on
document.write("You are using IE4 or higher");
@*/
I have not encountered these in production template code, and
the typed content change in CL 4962067 provides an escape-hatch
if conditional comments are needed.
R=nigeltao
CC=golang-dev
https://golang.org/cl/4999042
This simplifies transition functions to make it easier to reliably
elide comments in a later CL.
Before:
- transition functions are responsible for detecting special end tags.
After:
- the code to detect special end tags is done in one place.
We were relying on end tags being skipped which meant we were
not noticing comments inside script/style elements that contain no
substitutions.
This change means we will notice all such comments where necessary,
but stripTags will notice none since it does not need to. This speeds
up stripTags.
R=nigeltao
CC=golang-dev
https://golang.org/cl/5074041
The documentation for bytes.Replace says it copies
the slice but it won't necessarily copy them. Since
the data is mutable, breaking the contract is an issue.
We either have to fix this by making the copy at all
times, as suggested in this CL, or we should change the
documentation and perhaps make better use of the fact
it's fine to mutate the slice in place otherwise.
R=golang-dev, bradfitz, adg, rsc
CC=golang-dev
https://golang.org/cl/5081043
Use gobs to serialize indexes instead of encoding/binary.
Even with gobs, serialize data in slices instead of
applying gob to the entire data structure at once,
to reduce the amount of extra buffer memory needed
inside gob.
7x faster Write/Read for new BenchmarkSaveRestore
compared to old code; possibly because encoding/binary
is more expensive for int32 slice elements (interface
call to get little/big endian encoding), while gob's
encoding is fixed (unconfirmed).
new (using gobs):
suffixarray.BenchmarkSaveRestore 1 2153604000 ns/op
old (using encoding/binary):
suffixarray.BenchmarkSaveRestore 1 15118322000 ns/op
The actual serialized data is slightly larger then using
the old code for very large indices because full 32bit indices
require 5bytes using gobs instead of 4bytes (encoding/binary)
in serialized form.
R=r
CC=golang-dev
https://golang.org/cl/5087041
This CL generalises the pair of halfConnection members that the
serverConn holds into a single transport struct that is shared by
both Server and Client, see also CL 5037047.
This CL is a replacement for 5040046 which I closed by accident.
R=agl, bradfitz
CC=golang-dev
https://golang.org/cl/5075042