Remove all dependencies from non-test code in go/ssa to go/loader,
except the deprecated Create function which will be eliminated in
favor of ssautil.CreateProgram in a mechnanical followup.
Add Examples of two main use cases of SSA construction:
loading a complete program from source; and
building a single package, loading its dependencies from import data.
Add tests to ssautil of the two load functions.
Suggestions welcome for better names.
Planned follow-ups:
- replace all references to ssa.Create with ssautil.CreateProgram and eliminate it.
- eliminate support in go/loader for the ImportBinary flag, and the
PackageCreated hook which is no longer needed since clients can
create the package themselves (see Example).
Step 1 to fixing issue 9955.
Change-Id: I4e64df67fcd5b7f0c0388047e06cea247fddfec5
Reviewed-on: https://go-review.googlesource.com/8669
Reviewed-by: Robert Griesemer <gri@golang.org>
Missed this in a prior change.
Change-Id: I7358c17b73a1221cb8f9dff6b808fdea8b13ec06
Reviewed-on: https://go-review.googlesource.com/8916
Reviewed-by: Rob Pike <r@golang.org>
Depends on https://golang.org/cl/8767/.
With this change, cmd/vet does not depend on x/tools anymore
and could be moved into the std repo if so desired.
Change-Id: Ia205c6e1a6a63eebb27776064e5c24491043b683
Reviewed-on: https://go-review.googlesource.com/8791
Reviewed-by: Rob Pike <r@golang.org>
This package was only imported for the trivial Unparen function.
Change-Id: I0ead916a7fdb469a26b4fe99c6964a8ed1438c49
Reviewed-on: https://go-review.googlesource.com/8566
Reviewed-by: Rob Pike <r@golang.org>
Currently, if for some reason http.ListenAndServe fails, any running
running godoc processes don't get killed. I don't think this would ever
actually happen because, with godoc being set up in a separate go
routine, http.ListenAndServe would always(?) fail before the godoc
server started.
This change ensures that, if a Proxy has a cmd, it is closed when
http.ListenAndServe fails.
Change-Id: I0d3bfae0c16bc583248c2052a4d7a84c95127e76
Reviewed-on: https://go-review.googlesource.com/8570
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
...otherwise the interpreted program can make the whole test suite
slow. Just ignore the argument and return the current GOMAXPROCS
value.
Change-Id: Ife2ad6c53e6fdf9feea1d1b231d8d796b3db3a24
Also: add missing intrinsic for os.runtime_beforeExit.
Reviewed-on: https://go-review.googlesource.com/8591
Reviewed-by: Robert Griesemer <gri@golang.org>
Such errors include files without package decl, or conflicting package
decls. We ignore type errors, so we should ignore import errors too.
Fixes#10347
Change-Id: I9011a9099a2804281ea2d989d7263a9ce691be16
Reviewed-on: https://go-review.googlesource.com/8498
Reviewed-by: David Crawshaw <crawshaw@golang.org>
This package was only imported for the trivial Unparen function.
Change-Id: I14f8d91bc0afaa6ab3aa797a53e42e56b59ffcbe
Reviewed-on: https://go-review.googlesource.com/8499
Reviewed-by: Alan Donovan <adonovan@google.com>
This makes type-based queries faster and more robust because they
needn't invoke a C compiler, at the cost of worse results in/near .go
files that import "C". It's particularly important for "referrers"
since a refererrs query on a name from the standard library can load
every package in the workspace.
Fixes issue 10347
Change-Id: I2f65474ce963de5c0897ba67eeb26290dd449cf1
Reviewed-on: https://go-review.googlesource.com/8493
Reviewed-by: David Crawshaw <crawshaw@golang.org>
This is a putative fix for the file descriptor exhaustion problem
described in https://github.com/golang/go/issues/10306.
Change-Id: If603fb9bbaec1b53f6b44d15b2c202e4670035ce
Reviewed-on: https://go-review.googlesource.com/8421
Reviewed-by: Matt Joiner <anacrolix@gmail.com>
Reviewed-by: David Crawshaw <crawshaw@golang.org>
This peculiar case arose in range statements but there are other contexts
and one turned up in the auto-generated translation of the compiler.
Take care of it always.
for i := 0; i < 0; func() {i++; q=q.Link}() { ... }
That code has been given the obvious rewrite but we should still handle it.
Odd but easy to fix (tricky to test).
Fixes#10269.
Change-Id: I66e1404eb24da15a24be7f67403e19ed66fba0a7
Reviewed-on: https://go-review.googlesource.com/8284
Reviewed-by: Robert Griesemer <gri@golang.org>
Added test case. This required making the result sort order
deterministic when the results are spread across several packages.
Also: implements: print type names relative to query package.
Updated tests.
Change-Id: I9f882cd358a612585a4aac9a117b89d9131a294e
Reviewed-on: https://go-review.googlesource.com/8283
Reviewed-by: David Crawshaw <crawshaw@golang.org>
The check for len(argv)==0 now only applies to these modes.
Also, more consistent variable naming.
Change-Id: I9adb6bebc819eb43d54ddf63c42d952671ce9236
Reviewed-on: https://go-review.googlesource.com/8244
Reviewed-by: David Crawshaw <crawshaw@golang.org>
Features:
More robust: silently ignore type errors in modes that don't need
SSA form: describe, referrers, implements, freevars, description.
This makes the tool much more robust for everyday queries.
Less configuration: don't require a scope argument for all queries.
Only queries that do pointer analysis need it.
For the rest, the initial position is enough for
importQueryPackage to deduce the scope.
It now works for queries in GoFiles, TestGoFiles, or XTestGoFiles.
(It no longer works for ad-hoc main packages like
$GOROOT/src/net/http/triv.go)
More complete: "referrers" computes the scope automatically by
scanning the import graph of the entire workspace, using gorename's
refactor/importgraph package. This requires two passes at loading.
Faster: simplified start-up logic avoids unnecessary package loading
and SSA construction (a consequence of bad abstraction) in many
cases.
"callgraph": remove it. Unlike all the other commands it isn't
related to the current selection, and we have
golang.org/x/tools/cmdcallgraph now.
Internals:
Drop support for long-running clients (i.e., Pythia), since
godoc -analysis supports all the same features except "pointsto",
and precomputes all the results so latency is much lower.
Get rid of various unhelpful abstractions introduced to support
long-running clients. Expand out the set-up logic for each
subcommand. This is simpler, easier to read, and gives us more
control, at a small cost in duplication---the familiar story of
abstractions.
Discard PTA warnings. We weren't showing them (nor should we).
Split tests into separate directories (so that importgraph works).
Change-Id: I55d46b3ab33cdf7ac22436fcc2148fe04c901237
Reviewed-on: https://go-review.googlesource.com/8243
Reviewed-by: David Crawshaw <crawshaw@golang.org>
Emacs integration:
- eliminate oracle minor mode
- in go-mode, bind F5, F6 to "describe", "referrers".
This reverts a previous policy decision but convenience matters too.
- don't insist on an analysis scope for modes that don't do PTA.
- don't hide the filename as "▶"; show the last 20 chars.
(Especially useful for "referrers" mode.)
- output postprocessing: don't get stuck in a loop if the output
is not as expected (e.g. when it includes a panic log).
referrers:
- show the matching lines (like grep does).
We do the I/O in parallel.
Change-Id: I86b18c1d3a4d9fa4242984cba62b314796669d8e
Reviewed-on: https://go-review.googlesource.com/8120
Reviewed-by: David Crawshaw <crawshaw@golang.org>
AddImport and AddNamedImport attempt to place new
imports in roughly the correct place--and thus the
correct group--by matching prefixes. Matching prefixes
byte-by-byte led to "regexp" being grouped with "rsc.io/p".
Instead, match prefixes by segments.
Fixesgolang/go#9961.
Change-Id: I52b7c58a9a2fbe85c2b5297e50c87d409364bda3
Reviewed-on: https://go-review.googlesource.com/8090
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Cover deleted all comments because they can break the simple way that
counters are injected into the rewritten source. But //go: comments have
semantic value, and for instance go test -cover runtime fails during
compilation because of their absence from the annotated source.
We can keep the //go: comments because they are at the beginning of
the line and are not affected by our counter injection.
Fixes#10270.
After this CL, go test -cover runtime works.
A testing strategy that does not involve a golden file would be welcome
but I can't think of one.
Change-Id: I73f7b7a36383a8efed8e33fa2414cd0eac7d015a
Reviewed-on: https://go-review.googlesource.com/8173
Reviewed-by: Robert Griesemer <gri@golang.org>
Given
x()
panic(1)
y()
the y should not show as covered.
Fixes#10185
Change-Id: Iec61f1b096a888e6727be5f4526508654f5d3c91
Reviewed-on: https://go-review.googlesource.com/8140
Reviewed-by: Robert Griesemer <gri@golang.org>
Running 'godoc -src' would end up with concatenated sources:
$ godoc -src github.com/bradfitz/http2 Frame
// a frameParser parses a frame given its FrameHeader and payload
// bytes. The length of payload will always equal fh.Length (which
// might be 0).
type frameParser func(fh FrameHeader, payload []byte) (Frame, error)//
A Frame is the base interface implemented by all frame types.
// Callers will generally type-assert the specific frame type:
// *HeadersFrame, *SettingsFrame, *WindowUpdateFrame, etc.
//
// Frames are only valid until the next call to Framer.ReadFrame.
type Frame interface {
Unconditionnally insert two newlines:
$ godoc -src github.com/bradfitz/http2 Frame
// a frameParser parses a frame given its FrameHeader and payload
// bytes. The length of payload will always equal fh.Length (which
// might be 0).
type frameParser func(fh FrameHeader, payload []byte) (Frame, error)
// A Frame is the base interface implemented by all frame types.
// Callers will generally type-assert the specific frame type:
// *HeadersFrame, *SettingsFrame, *WindowUpdateFrame, etc.
//
// Frames are only valid until the next call to Framer.ReadFrame.
type Frame interface {
Fixes#9726.
Change-Id: I51ee04e53d4962c890ea601125eda8fce84d6a46
Reviewed-on: https://go-review.googlesource.com/7681
Reviewed-by: Robert Griesemer <gri@golang.org>
See golang.org/cl/4131 for context.
Change-Id: I474d4d455ad0207bf6cea6c36fc8bdcea3405b47
Reviewed-on: https://go-review.googlesource.com/7726
Reviewed-by: David Crawshaw <crawshaw@golang.org>
Silly test added yesterday requires that some code in a goroutine executes.
Make sure it does.
Change-Id: I7e852454736e300151473986cc437a70b41dc9b7
Reviewed-on: https://go-review.googlesource.com/7691
Reviewed-by: Russ Cox <rsc@golang.org>
This was not a visible bug since the only caller discards the relevant
result, so I also deleted the result.
Fixes#9999
Change-Id: I276d6523b2891d3cb9c8137448e1aed32a5fd197
Reviewed-on: https://go-review.googlesource.com/5921
Reviewed-by: Michael Matloob <michaelmatloob@gmail.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
This avoids littering the tree, and confusing some editors (e.g. Atom)
that expect the inode number to remain constant.
Change-Id: I2faeda1ed1b01e5e4cc720744ea3c99ab29e7333
Reviewed-on: https://go-review.googlesource.com/7664
Reviewed-by: Robert Griesemer <gri@golang.org>
eg with no arguments prints its usage, including a hand-written flag summary.
eg -help shows the detailed help message.
Change-Id: I615d8de3985ced1e86e9d7cafa9ef679079b249c
Reviewed-on: https://go-review.googlesource.com/6951
Reviewed-by: Robert Griesemer <gri@golang.org>
Composite literals are initialized in place where possible, but in
cases the initializer expression refers to the variable that
is being updated
x = T{a: x.a}
we must ensure that the RHS is fully evaluated before we execute any
stores to x. This means we need to record the sequence of stores in a
"store buffer" and execute it only once the entire composite literal
has been evaluated.
Fixes issue #10127
Change-Id: If94e3b179beb25feea5b298ed43de6a199aaf347
Reviewed-on: https://go-review.googlesource.com/7533
Reviewed-by: Robert Griesemer <gri@golang.org>
It used to do packages only when run by the go tool, but it was
fixed a while back to handle packages properly when doing a
directory walk. Remove the incorrect information from the
documentation.
Change-Id: I961340bb84e48474c94ee03bf88f9136492c0226
Reviewed-on: https://go-review.googlesource.com/7642
Reviewed-by: Andrew Gerrand <adg@golang.org>
Just missed a case (ha!) in the tree walk. Dup the code for an empty switch, add test.
Fixes#10163.
Change-Id: I3d50ab6cb450ca21e87213291eaab8cbe924fac5
Reviewed-on: https://go-review.googlesource.com/7641
Reviewed-by: Andrew Gerrand <adg@golang.org>
Skip tests on arm platforms.
The godoc tests require large amounts of memory, in excess of 700mb in -index mode which none of the arm builders have spare.
Because of their requirements the tests can be killed by the test runner leaving stray godoc processes spinning in swap trying to -index.
Change-Id: I1544d56e9d9aabbbaac21adeebfb9e2690bd2da5
Reviewed-on: https://go-review.googlesource.com/7540
Reviewed-by: Andrew Gerrand <adg@golang.org>
Be more tolerant in the presence of incorrect range clauses
if for loops and type-check loop body with minimal assumptions
about iteration variables. (Before, in some cases we would simply
ignore the loop body in such cases).
Fixes#10148.
Change-Id: I0b66f81875348088c1a7fa04ccdcbfe768f2eb6c
Reviewed-on: https://go-review.googlesource.com/7525
Reviewed-by: Alan Donovan <adonovan@google.com>
Since all SSA values are immutable, no value copying is required for
any operations except those that load from or store to a variable;
those operations must do an aggregate copy, i.e., descend into struct
and array elements. All other calls to copyVal have been removed;
they were pieces of duct tape, as I had long suspected.
The descent must be based on the static type information, not the
"shape" of the dynamic value, since two reflect.Value structs may have
different internal shapes. We clobber the true definition of
reflect.Value's underlying type, replacing it with struct{interface{},
interface{}}, which is close enough to make the load/store functions
work.
+ Test
Change-Id: I5e239d91ed0cb2a669a9f75766024fe1f9a5c347
Reviewed-on: https://go-review.googlesource.com/7532
Reviewed-by: Robert Griesemer <gri@golang.org>
Config.Cwd sets the base directory; os.Getwd is its default.
+Test.
Change-Id: I213abfb30085cd1306719ed6f94aeae6a3170bc0
Reviewed-on: https://go-review.googlesource.com/7502
Reviewed-by: Robert Griesemer <gri@golang.org>
Sorry for the oversight.
Change-Id: Ibb686dbee996b5223bd223fdd3afaab243a7a3ee
Reviewed-on: https://go-review.googlesource.com/7501
Reviewed-by: Robert Griesemer <gri@golang.org>
On tip, search included redundant source results from /pkg/bootstrap
(with broken links as godoc doesn't support source files under /pkg).
This change excludes all directories under /pkg from indexing.
Fixesgolang/go#10024.
Change-Id: I0c69d22ff08d131f9c37c91a7711db6a4ec53fd4
Reviewed-on: https://go-review.googlesource.com/7267
Reviewed-by: Andrew Gerrand <adg@golang.org>
Also use a more reliable implementation.
Change-Id: I9e6858c7e9bdb60f1fb4e060e6d4d1b3762b83bc
Reviewed-on: https://go-review.googlesource.com/7260
Reviewed-by: Minux Ma <minux@golang.org>
Run godoc indexing just once on startup. Wait for indexing to complete
before switching to new side. Increase startup timeout to accommodate for
indexing.
Updates golang/go#9996.
Change-Id: I1e746a68b7d787e6d7f180c2617ea75f0d3291f8
Reviewed-on: https://go-review.googlesource.com/7120
Reviewed-by: Andrew Gerrand <adg@golang.org>
The caller of Usage should call os.Exit -- Usage shouldn't call it.
Change-Id: I3decf662883fb2a6b19b7035138ee8a06a02de08
Reviewed-on: https://go-review.googlesource.com/7110
Reviewed-by: Andrew Gerrand <adg@golang.org>
The callbacks are intentionally concurrent, making this function very
easy to misuse (most clients so far have got it wrong, even my own).
Using a channel in the API makes the concurrency obvious, the
correct usage easy, and the client control flow simpler.
Change-Id: Ied38c3ed5c98b40eb1b322a984ed9dc092ac0918
Reviewed-on: https://go-review.googlesource.com/3250
Reviewed-by: Sameer Ajmani <sameer@golang.org>
Before this change, declared init functions were not package members;
this choice dates from when go/types did not create Func objects for them.
Now, they have an Object. They appear in Members, keyed by "init#%d"
(sequence number) for uniqueness. They can be enumerated. They can
be looked up from a *types.Func via (*Program).FuncValue.
Caveat: fn.Object.Name() no longer equals fn.Name() in all cases.
NB: incompatible API change! (Your build will not break though.)
Change-Id: I2de873079fd57329e6c2f55a282940f6699a77a1
Reviewed-on: https://go-review.googlesource.com/6950
Reviewed-by: Robert Griesemer <gri@golang.org>
Reviewed-by: Peter Collingbourne <pcc@google.com>
