Narrow down parsers based on option string
Instead of trying to match all patterns one by one, we can check the
start of the option string to be parsed and narrow down which pattern
(parsers) should be used.
Doing so is much less elegant code-wise as we are "unrolling" the
beginning of the patterns by hand, but it gives a good speed up. Around
300ms saved with a test of LLVM code base (~50% of parseOption time) and
much larger gains on much larger projects or course.
Change-Id: I9e841e7233078d6e38ad08943d98966d0e3c661e
Signed-off-by: Marc-Andre Laperle <malaperle@gmail.com>
Make sure the project is still accessible before doing the work.
It's possible that the project got closed before the job had a
chance to start.
Change-Id: If02e818394502397e8958d18284160d4f08ca195
Signed-off-by: Marc-Andre Laperle <malaperle@gmail.com>
Optimize AbstractLanguageSettingsOutputScanner.findResource by checking
under the currentProject first when searching for the IFile
corresponding to the parsed file name.
When the parsed file name is absolute, we can try to make it relative to
the current project and see if it exists as an IFile instead of going
through findFilesForLocationURI which is very slow. Having an absolute
path for the source file is quite common under the current (local)
projet.
This saves around 1.5 sec when parsing all commands of LLVM projet as an
example.
Change-Id: I576a917410e5d5ecbd8e932011555c699b250354
Signed-off-by: Marc-Andre Laperle <malaperle@gmail.com>
- fix Core Make Makefile template to add one for linuxtools
build mode which is used for Linux Tools profiling
- fix the error default to change tab indent to spaces so
make won't treat error message as recipe
- add some default profiling flags for gcov and gprof support when
Linux Tools profiling
Change-Id: I29c2247c9e2b00a4e1c1952198faf881abd7a66a
This make messages stand out more prominently in the build console.
Change-Id: I180a7eb6631870004cca94a7f1d2c8bda5b044f8
Signed-off-by: Martin Weber <fifteenknots505@gmail.com>
And prepare to make it an error in CDT to not have properly handled
an Autocloseable which means a number of fixes to make sure handles
are closed.
Change-Id: I36cd46017bbce6ece1703d688d7754e523eca68f
buildInfo is null in the case that the node list is empty.
Change-Id: I165ec31eeeab57e1eb7bed0268d0f83c70d6e0f7
Signed-off-by: Torbjörn Svensson <azoff@svenskalinuxforeningen.se>
Removed useless catch block that just rethrows the same exception
Change-Id: If3e4faea1ed889d4f7b39a7cff786e0aa9a6803d
Signed-off-by: Torbjörn Svensson <azoff@svenskalinuxforeningen.se>
Compile a pattern ahead of time and early return.
The first pattern is used to trim extra characters that are not
contained within the actual option pattern. If this first pattern
doesn't match, there is no point in continuing because it is a superset.
Times before/after, only counting AbstractOptionParser.parseOption
LLVM before: 4289ms, after: 622ms
Change-Id: Id40fc9a35359c39aea00ba14813ffe6c343158fc
Signed-off-by: Marc-Andre Laperle <malaperle@gmail.com>
Clarify in the javadoc that numbered back-references should not be used
in order for option pattern to safely work. Add a unit test
demonstrating the flaw when numbered back-references are used. Update
GCC parser as an example for not using numbered back-references.
Change-Id: I008b3589486dc9fb0d9d9aa41f7cc2443bf4351e
Signed-off-by: Marc-Andre Laperle <malaperle@gmail.com>
Remove all equalIgnoreCase and equal with uppercasing for environment
variables
Change-Id: Ic15974b5fb62413c7b1826ced544ff6d4a8eba2f
Signed-off-by: jantje <eclipse@baeyens.it>
Unused since bug 559674: Integrate new indexer support into CDT build
Change-Id: Ifd4878f944eff6f12e3ab2df399c619abfa47f59
Signed-off-by: Martin Weber <fifteenknots505@gmail.com>
Sets the pattern to the default of upcoming tycho 2.0 in advance.
Signed-off-by: Martin Weber <fifteenknots505@gmail.com>
Change-Id: I31b3fc733d0cb888fbf6f566995ce2043f6cd621
The vendor specific support files need to be installed separately.
Includes changing the bundles display names to match the CMake main bundle
so that in the install wizard it is clear(er) what they work with.
Change-Id: I12ca155228e906c6c38fe37e37e8ce08d05a2452
This is one way to publish the cmake.is plug-ins. It may be that
additional features are desired as this implementation is all of cmake
support or non of it.
Change-Id: Ie945d2ce94b5ac34c6c238ed7bfbdc3ff336e538
Implement a file exclusion algorithm that favors excluding whole folders when
possible.
The way it works is we gather exclusion information of each folder as we visit
each children. When "leaving" the folder, we can act on whether or not it can
be considered for exclusion as a whole or instead individually exclude a subset
of its children.
Using LLVM code base as a test:
Before: 613 sec
After: 2.4 sec
Change-Id: Ib882a72cae157e3db6b6c94a1a09cb6f05b66bc4
Signed-off-by: Marc-Andre Laperle <malaperle@gmail.com>
Cache results of various path resolution algorithms.
Resolving paths is particularly slow while creating entries, see
AbstractLanguageSettingsOutputScanner.createResolvedPathEntry.
There are three main callees within that method that this patch addresses with
a caching approach:
* findContainerForLocationURI: First, this finds containers for a given URI in
the workspace using Eclipse resources API. Then a single container is
selected based on a preferred project. This can done repeatedly for include
paths, which are often similar for source files in a given project or source
folder. This first step is the expensive one and it only depends on one
argument (the URI) and a simple IResource[] return type, so the cache here is
done for this operation. Then the post-filtering is kept as is.
* findFileForLocationURI: Similar to the container case but for files. A
typical projet has much less file paths than folder paths in its options. One
more common option using file paths is -include. The same approach is applied
here as the previous point because there are performance gains but they are
smaller if you consider typical projet setup.
* findBestFitInWorkspace: When a path cannot be found, this makes an attempt to
find the parsed path relative to every folder of the workspace, by starting
first with the preferred project, then its referenced projects and then the
rest. Caching the result of findBestFitInWorkspace itself is too cumbersome
because the result depends on 3 variables (currentProject,
currentCfgDescription and parsedName) which would make a complex cache key.
Instead, caching the result of findPathInFolder at the project level is
sufficient, with little to no performance difference.
In all three cases, the class LRUCache is used in order to limit memory
consumption of the cache. A limit of 100 elements for each cache was chosen
based on experimentation with a few projects like LLVM and projets several
times bigger. A limit higher than necessary for small projects does not incur a
noticeable overhead for small projects and a limit too small for very large
projects merely diminishes the performance gains.
Using LLVM code base as a test, the time to parse options for all files:
Before: 68395ms, after: 5599ms
Change-Id: Ib997e9373087950f9ae6d93bbb1a5f265431c6bc
Signed-off-by: Marc-Andre Laperle <malaperle@gmail.com>
By design, the parser/provider fires when loading the cproject
description but we also call it by hand in the main code of the tests.
This means CompilationDatabaseParser could be running twice
simultaenously along with the same output parser code that it delegates
to. The problem was exposed fully when adding more complex data
structure (hash maps) to the output parsers in another commit, as it
would produce ConcurrentModificationException.
We need to be careful when we choose to call setProjectDescription
because it triggers the provider and then we have to wait for the jobs
to complete (joinLanguageSettingsJobs). By taking this into
consideration, several tests had to be updated. Most notably, the
read-only config test case had to be merged with the time-stamp update
test case because it was the only sensible way to test before/after
changes of language settings with such config.
Change-Id: Ib3a7caefa95b436ad9b699a2614e966a4a8dfca9
Signed-off-by: Marc-Andre Laperle <malaperle@gmail.com>
Don't overwrite it in plugins as version is set in
parent/pluginManagement.
Change-Id: I73d2b4d234ba83eae7ec2cd51f3e53d58256b81e
Signed-off-by: Alexander Kurtakov <akurtako@redhat.com>
Get rid of tycho-extras-version as there is no more tycho-extras.
Change-Id: I9d215aad94c4e8320153f06368d6b3ecfa4a3c42
Signed-off-by: Alexander Kurtakov <akurtako@redhat.com>
One flaw with this implementation is that the "arguments" coming from
the CDB do not have shell quoting and shell escaping of quotes whereas
the current implementations of Build Output parsers assume some form of
shell quoting. This means that simply joining strings of arguments with
spaces will be missing the expected shell quoting and possibly misparsed
by the build output parsers.
It is not clear to be at this point if this should be fixed or not as it
might involve revamping the existing build output parsers to add the
concept of shell/environment and this could also affect potential
extenders.
In this current form, simple cases with no spacing and quote escaping
involved work correctly and is still a nice improvement.
Change-Id: Ia81796e63c748318b34696998ac4a467712e5f96
Signed-off-by: Marc-Andre Laperle <malaperle@gmail.com>
Remove some Wizard classes that are not referenced anywhere anymore.
This commit is just a first of probably several commits but is already
a start. I already had very large commits in progress in the past but
it became big and hard to push so I'd rather do it step by step this
time and at least have some of it done for the next release.
Also moved some messages to its own message bundle in managedbuilder.ui
because it's the only place they are used now.
Change-Id: Ib4258684c91f205dc4af3b17169609b5ebcff253
Signed-off-by: Marc-Andre Laperle <malaperle@gmail.com>
The dependency was removed.The Copyright head was updated.
Change-Id: I82e4289fc8cb14e8488b847e8efa415dd1d77b6c
Signed-off-by: Sergei Kovalchuk <serjiokov@gmail.com>
The class overrode org.eclipse.core.runtime.Platform to
workaround bugs in the platform that have since been fixed.
As 32-bit x86 and PPC support has been removed this
class is no longer needed as all the code is now
unreachable anyway.
Change-Id: I01bb00b9203aa02663ff25ce36c4c14f22dadee5
Signed-off-by: jantje <eclipse@baeyens.it>
The method Configuration.getUserObjects() and Configuration.getLibs()
should respect the commandGenerator and applicabilityCalculator defined
for the option in the toolchain definition.
The method Tool.getToolCommandFlags() should call the commandGenerator
only if the value type would generate a default command.
Change-Id: I4d6224627888e602682076ac929adbf808cff8d7
Signed-off-by: Torbjörn Svensson <azoff@svenskalinuxforeningen.se>
Use java.text.MessageFormat instead of com.ibm.icu.text.MessageFormat
Change-Id: Ie7ed91d776b707def1bbfc12c24f5e863638e083
Signed-off-by: Sergei Kovalchuk <serjiokov@gmail.com>
For example, you can use ${ProjDirPath}/buid/compile_commands.json
which is a more sharable setting
Change-Id: Ic51775ea1f7d258c55e2e720c94f626a17444744
Signed-off-by: Marc-Andre Laperle <malaperle@gmail.com>
