Summary:
This makes sure that the javac_jar is disabled when setting the external compiler option to point at the Infer wrapper
Closes#976
Reviewed By: jvillard
Differential Revision: D9193336
fbshipit-source-id: abafb51fc
Summary:
Forcing integration with `--force-integration` would only work for some
integrations that stayed in OCaml-land. This propagetes the forcing to
infer.py.
Also remove some unused options on the Python side, and add more debug
information.
Fixes#927
Reviewed By: mbouaziz
Differential Revision: D8235504
fbshipit-source-id: 1d98543
Summary:
Change the license of the source code from BSD + PATENTS to MIT.
Change `checkCopyright` to reflect the new license and learn some new file
types.
Generated with:
```
git grep BSD | xargs -n 1 ./scripts/checkCopyright -i
```
Reviewed By: jeremydubreil, mbouaziz, jberdine
Differential Revision: D8071249
fbshipit-source-id: 97ca23a
Summary: Fixing the support for single core analysis in 8ce79a0613 was improving the performance of the genrule-based integration by more that 2X. Running Infer in single core more with the external compiler integration for Buck also improves the performance significantly.
Reviewed By: mbouaziz
Differential Revision: D7989255
fbshipit-source-id: 71fb842
Summary:
It's already turned of systematically for the Java integration, this just
generalises it. The Buck daemon seems to cause issues with infer from time to
time that are hard to debug.
Reviewed By: jeremydubreil
Differential Revision: D7400068
fbshipit-source-id: f05ee07
Summary:
This is to fix the conflicts between Eradicate and the Biabduction when reporting the same kind of errors: when Eradicate is on, the Eradicate warnings will have priority over the null deference reported by the biabduction.
If this approach proved to be successful in prod, I will refactor the reporting mechanism in the analysis itself to simply not report the null dereference in this case at all. For the codebases that aren't yet fully consistently using `Nullable`, this combined approach looks like a good way to deploy Infer toward full null safety.
Reviewed By: mbouaziz
Differential Revision: D7102119
fbshipit-source-id: 35d3add
Summary:
In buck queries, the `regex` in `kind(regex, ...)` is open, it can match superstrings, like `prebuilt_cxx_library` when we want `cxx_library` instead.
This adds `^` and `$` to our existing kind filtering.
Also optimizes the query in `buck_target_determinator.py`.
Reviewed By: dulmarod
Differential Revision: D6722665
fbshipit-source-id: 22d839f
Summary: A source can belong to more than one target. In this case, we should keep only one of the report.
Reviewed By: sblackshear
Differential Revision: D6200058
fbshipit-source-id: 4eced42
Summary: Saving the list of bugs in a set removes the ordering. Also, there should be no need to remove the duplicated warnings at this level.
Reviewed By: sblackshear
Differential Revision: D6060554
fbshipit-source-id: a78d35d
Summary: That was too noisy. Propagate `--quiet` to the Python reporting hook so as to still emit bugs.txt and so on.
Reviewed By: martinoluca
Differential Revision: D5501106
fbshipit-source-id: 63b6451
Summary: The Buck cache was not correctly invalidated when switching between the different analysis mode. This was causing the analysis results to be mixed up. This revision should fix this.
Reviewed By: sblackshear
Differential Revision: D5073606
fbshipit-source-id: eb14418
Summary:
This option is not needed anymore as it was introduced to counter an uncovered
perf issue with creating human readable reports. The perf issue has been
addressed.
Instead of this option, one can use `infer --report-hook /bin/true ...` to
disable reporting. However, right now the Buck integration doesn't honor it so
this would need to be fixed to be a true equivalent of `--disable-bug-list`.
Reviewed By: jberdine
Differential Revision: D4712877
fbshipit-source-id: a09304f
Summary: The Buck integration assumes that the output jar file of all the dependency targets are available locally in order to retrieve the analysis from these targets. However, this is not guaranteed to be true when there is a cache hit on some targets. Running `buck build` with the option `--deep` forces this property.
Reviewed By: jvillard
Differential Revision: D4474036
fbshipit-source-id: accabfa
Summary:
Also make sure we don't introduce deprecated options in our repo, eg when
calling infer from infer.
Reviewed By: jeremydubreil
Differential Revision: D4430379
fbshipit-source-id: 77ea7fd
Summary:
This will help porting more Python code over to OCaml. Since the reporting step
uses a lot of Python libraries that would be a pain to rewrite in OCaml (eg,
syntax highlighting), keep this functionality in Python and make it possible to
call it from OCaml as a script.
Reviewed By: jberdine
Differential Revision: D4182832
fbshipit-source-id: fc83220
Summary:
This makes our python code work (instead of crashing) when the source file
should be found not from the current directory (or absolute path), eg with
`infer --project-root .. -- clang -c hello.c`.
Reviewed By: jeremydubreil
Differential Revision: D4130802
fbshipit-source-id: 001f72d
Summary:
During the incremental analysis using the Buck distributed cache, if there is a cache hit for a given module, the output jars for the intermediate targets are not necessarily dowloaded. We therefore need to filter the jar files that are present on disk before loading the analysis artifacts from it.
This will also be neccesary when combined with the --keep-going option of Buck
Reviewed By: sblackshear
Differential Revision: D3941853
fbshipit-source-id: befda63
Summary:
Before this diff, Infer was simply going through the list of jar files found in `buck-out` and was loading all the `report.json` files found in those jar files in order to merge them into a final report. The main drawback of this was that removing `buck-out` was mandatory to get accurate results when switching between targets to analyze.
With this diff, we now use the `buck audit classpath` option to get from a list of targets, the list of jar files to load the `report.json` files from. This allows to more easily use Infer from the command line when switch branches from the repositories or when switching between targets to analyze.
Reviewed By: martinoluca
Differential Revision: D3922548
fbshipit-source-id: ec550fa
Summary: The reactive mode is not currently supported for Java Buck projects. Print a clear error message to avoid confusion. Using `--reactive` with Buck projects is crashing. When using reactive, the backend crashes because it tries to find the `.start` file in `infer-out` whereas Infer saves the analysis artifacts next to the generated classes following the `-d` options with Buck so that they can be cached. Pointing the backend to the right Infer out would not help as this directory will be blown up as soon as the files are modified.
Reviewed By: sblackshear
Differential Revision: D3901005
fbshipit-source-id: d8dcbce
Summary: We no longer need to disable the Buck daemon since we not pass the path to the compiler via the option `--config` instead of modifying the .buckconfig.local file
Reviewed By: sblackshear
Differential Revision: D3854109
fbshipit-source-id: 1126159
Summary:
This diff changes the toplevel 'infer' executable from the current
python script to an OCaml binary. Currently this executable only parses
command line arguments, sets up environment variables, and invokes the
existing python script. This improves infer's command-line and
configuration interface, since passing arguments to the frontends or
backend no longer requires manually setting environment variables, and
arguments for the toplevel can now also be specified in .inferconfig.
Simplification and migration of functionality from the python script is
left for the future.
Reviewed By: martinoluca, jvillard
Differential Revision: D3450662
fbshipit-source-id: 1b52302
Jeremy Dubreil
Martino Luca
Jules Villard
Tony Hu
Mehdi Bouaziz
Sam Blackshear
Josh Berdine
Dulma Churchill