Add top of file documentation for crashcontext.mli

Reviewed By: jeremydubreil

Differential Revision: D3826480

fbshipit-source-id: fa17ae5

infer/src/backend/crashcontext.mli

@@ -7,4 +7,36 @@
  * of patent rights can be found in the PATENTS file in the same directory.
  *)
 
+(**
+   CRASHCONTEXT Intro [experimental]:
+
+   Crashcontext is an experimental analysis (in the future: a family of
+   analyses). It takes one or more stacktraces representing crashes
+   corresponding to the codebase being analyzed and expands them into
+   crashcontext.json files. These files incorporate further information about
+   the code that might have executed before the crash.
+
+   This analysis is run with '-a crashcontext' and must take either of the
+   following extra arguments:
+
+    --stacktrace stacktrace.json (a single stacktrace, output defaults to
+                                  crashcontext/crashcontext.json)
+    --stacktraces-dir dir (will expand every json stacktace in dir, output
+                           crashcontext files will be saved to dir as well)
+
+   For further information, take a look at tests under:
+
+    infer/tests/codetoanalyze/java/crashcontext/ and
+    infer/tests/endtoend/java/crashcontext/
+*)
+
+(**
+   Runs crashcontext epilogue code, which takes the per-method summaries
+   produced by crashcontext related analysis (future: analyses) and stitches
+   them together into a final crashcontext.json output file.
+   This code should run after all checkers when running with '-a crashcontext'.
+   When running with buck, summaries are stitched across multiple buck targets,
+   so this runs at the end of the parent buck infer process only.
+   TODO: Similar integration with build systems other than buck.
+*)
 val crashcontext_epilogue : in_buck_mode:bool -> unit