The extra nodes in .expected files are due to the changes from
https://github.com/github/codeql/pull/13717, which are not applied to
configuration classes extending DataFlow::Configuration or
TaintTracking::Configuration.
The new `edges` and `nodes` sections in the .expected files are because
the PathGraph module was not imported in the tests before, and thus
these query predicates were not in scope.
The extra nodes in .expected files are due to the changes from
https://github.com/github/codeql/pull/13717, which are not applied to
configuration classes extending DataFlow::Configuration or
TaintTracking::Configuration.
We now precisely track flow from positional arguments to splat
parameters, provided that splat arguments are not used and there are no
positional parameters after the splat parameter. For example, in this
case:
def f(x, y, *z); end
f(a, b, c, d)
we get flow from `c` to `z[0]` and `d` to `z[1]`.
We get false flow if there are positional parameters after the splat
parameter. For example in this case:
def g(x, y, *z, w); end
g(a, b, c, d)
we get flow from `d` to `z[0]` instead of `w`.
We also track flow in this case
def f(a, *b)
sink b[0]
end
f(1, *[taint, 2])
Although this is actually using taint-tracking (so we can use the +=
statement), I would personally forget to check under the
dataflow/tainttracking folder to look for such a test, so I'm opting to
keep it under the dataflow/ folder.
For these tests, we cannot use the same mechanism, as we want the
data extensions to be available for both tests.
Instead, we create a ql-pack for the test directory and point to
the data entensions from there. This makes the extensions
available for all tests in the directory.
For this test, we can simply use the convention,
that a file called `[ql-file-stem].ext.yml` will be used
as data extensions exactly for the test represented by `ql-file`.
