In case the primary builder wants to depend on a go binary by name,
expose an interface to find the real installed path of the binary.
Most of this change is changing it so that we're storing the install
path without any ninja references.
Change-Id: I873378ebdd47d8036d3cef5aab76f036846a3db1
This should prevent the running of Travis tests from causing
a filesystem loop src.test/test_tree/blueprint -> ../..
Bug: 63100441
Test: NINJA=... ./tests/test_tree_tests.sh; stat src.test 2>/dev/null \
&& echo "src.test should not exist"
Change-Id: Ie01fe8999c738dca166e3fabde9d9391f6b70210
Bug: 64600838
Test: mkdir errtest \
&& ln -s /tmp/dontexist errtest/Android.bp \
# and add errtest to ./Android.bp \
&& m nothing \
# and check that the error message mentions a symlink
Change-Id: I841ec12d613f61ccc3396538062bee48c8c1ca27
This is necessary to support other go code that uses build tags to
control which files build on which platforms. For example, the
github.com/golang/protobuf/proto package includes two implementations of
pointer:
pointer_reflect:
// ...
// +build appengine js
// This file contains an implementation of proto field accesses using package reflect.
// It is slower than the code in pointer_unsafe.go but it avoids package unsafe and can
// be used on App Engine.
pointer_unsafe.go:
// ...
// +build !appengine,!js
// This file contains the implementation of the proto field accesses using package unsafe.
Without this change, microfactory tries to include both files, which
breaks, since that enables multiple implementations of the same symbols.
This implements the logic to find the +build comments (following the
spirit of the functionality in go/build, even though they make different
choices of parsers), and the logic to match the tags to the current
system (which isn't exported as public API, so we have a simplified
version).
Since we're only parsing the comments before the import statement, this
does not significantly affect performance.
Test: Build github.com/golang/protobuf/proto
Change-Id: I10c48472d3f9f3e96b800a5d7c15450b1737002b
In addition to running with `go run` and creating a microfactory binary,
allow microfactory to be used as a package from other go tools as well.
To allow other packages to use this, it needs to be in a non-main
package, but `go run` requires a main package. So microfactory.bash runs
a sed script before running microfactory with `go run`.
This could also be solved by using a relative import, but neither
blueprint nor microfactory currently support that.
Change-Id: I084163b14720102b3fb93a3c9d44b5d0225ff2c8
Propagate extra ninja file deps through mutators so that they
can use ctx.AddNinjaFileDeps.
Test: blueprint tests
Change-Id: I299a0665c3f63b020ae345889fd78b91b91b215a
Allow a module to create other modules as if they were specified
in a blueprint file. CreateModule takes the name of a module type,
calls the factory function the module type, and then inserts
properties into the property structs returned by the factory.
Bug: 64161749
Test: TestCreateModule
Change-Id: Ic1903305d6b08eae1edc7f0fb4137e85544aac0f
Blueprint was using "replace" semantics when unpacking properties
into property structs, meaning if a module factory pre-set property
values they would be overwritten by whatever was in the Blueprint
file. This is different than what would happen if the same property
was updated using the Append*Properties functions in proptools, which
would use "append" semantics, which append strings and lists,
logically ORs booleans and replaces pointers to strings and booleans.
Replace unpack's semantics with append semantics for consistency.
Any previous users of pre-set properties can move to using a pointer
to a string or boolean if they want the old behavior.
Test: unpack_test.go
Test: extend_test.go
Change-Id: I02eebe80916e578938142f8e76889bd985223afc
The next patch will introduce creating new modules, so rename the
current variable that contains new variations of existing modules
to avoid confusion.
Test: blueprint tests
Change-Id: Ic1d3824e54e51c282a08f8ecef8fc658cc503a65
This appears to be fixed in gimme 1.2.0, but that's not available in
Travis yet (even in the latest trusty images). This is the gimme change
to always set GOROOT:
913dd17ff3
Change-Id: Ic9024b4bbbf5c75ed162a03fe119e4db9a1cdbba
Otherwise they'll fall back to the GOROOT embedded in the distribution,
which may not be correct.
This used to be propagated through build.ninja.in, but exporting it in
the environment should catch any other Go users too.
Change-Id: I3d6de8494b266e9b127c0a528d31df03ddc5c3a0
Having separate .ninja_log files doesn't work very well after
04d886b110
since the last successful timestamp is not recorded in the ninja log.
This was triggering the primary builder to be run twice on every build
since it was always considered out of date in the primary, then main
stages.
The two bootstrap stages already were unified when minibootstrap was
simplified, but this combines the main stage as well. This means that
we'll save some time running globs that are shared between minibp and
the primary builder as well (subdirs globs).
The cleanup code needed to be refactored, since the ninja_log will have
entries from the main stage as well. So instead of looking at everything
in the ninja log, take a string prefix to limit what we may clean.
Change-Id: I8d43168643aa8a0c6c3e48c6101eaa45d174bbe5
Now that packages and binaries are only built when needed, they need to
be defined in both stages in case something building in the main stage
needs something that was defined, but didn't need to be built in the
earlier stage.
Change-Id: I67704fba1d670e82218eba5ce91f0c01896557a2
Instead of forcing things to be built alongside the primary builder,
mark all packages and binaries as optional, unless they're explicitly
marked as default.
Change-Id: Ie53c90a01b41ce886d651f9aa4c4900d1edb0157
This duplicates building common blueprint go packages between minibp and
the primary builder, but drastically simplifies the first stage,
removing the need to check in a generated build.ninja.in.
Change-Id: I639a9637f1ed36d4210823ef276c0f7a064a83bd
From commit 38cef8ac3979da8e13037540981fe7aecc1e1fbe
Removes Go 1.7 from travis, since this depends on os.Executable
Change-Id: I58e2ae542621ffad66e9e145031091cb93ebb497
If tests have a TestMain method the os import is unused and
breaks building the test.
Test: gotestmain/testmain_test.go
Change-Id: Ia46cdd0df71b0fc7a53d08644d220ecfd779b2ff
Use testing.MainStart to support test packages that implement
TestMain for initial setup and teardown.
Change-Id: I7330a8e922448063f291f7b5b2eed31d274248bd
Patterns that were not wild would return an empty "dirs" list if the
file was found. But then if they were removed, we wouldn't know to
update the glob file and re-run the primary builder.
In this case, instead of adding the final directory into the dirs list,
add the matched files themselves. Due to editors performing atomic
writes (the directory timestamp often gets updated at the same time as
file timestamp) this is probably more efficient. In either case, we're
only re-running the individual glob, which is rather cheap.
Rename startGlob/Glob return name from "dirs" to "deps" since it may
contain files now too.
Darwin has a default limit of 256 open files per process. Parsing
too many blueprint files in parallel can hit the limit. Cap the
concurrency at 200.
Test: manual testing with limit set to 32
Change-Id: Ic64d21d2c0ffd7c86bf3f02fb51216ee5684a80c
Spaces normally don't need to be escaped, but leading spaces are
trimmed. Escape leading space to allow setting a variable to a
value with leading spaces.
Test: ninja_string_test.go
Change-Id: Ic0ffb076dbd603b7c0203720b9c1ea635c5ded75
