Update shared data-flow libraries #241
Conversation
max-schaefer
force-pushed
the
update-data-flow
branch
from
73c1b6e to
dc5813b
Compare
owen-mc
left a comment
owen-mc
left a comment
There was a problem hiding this comment.
I see there are some slight differences between our dataflow libraries and the ones in the codeql repo. Why is that?
Which differences do you mean? |
| result = t // stub implementation | ||
| } | ||
| /** Gets the type of `n` used for type pruning. */ | ||
| DataFlowType getNodeType(Node n) { result = n.getType() } |
There was a problem hiding this comment.
For example, we don't have SuppressUnusedNode(n) here, cf https://github.com/github/codeql/pull/3854/files#diff-f034a858ef09d2a8a0060f8297421736R228
There was a problem hiding this comment.
Right; those are the language-specific bits. Basically, any file that starts with import go is language-specific, and while it will look very similar to the corresponding files for other languages, there are many minor differences. In this case, C++ doesn't use the node type to restrict flow, so they have to use this trick to avoid getting warnings from the QL compiler.
Anders recently added an excellent overview of the data-flow libraries for language maintainers which you can consult for more details.
Mention Go cookbook queries in 1.24 change notes
2 participants
It's been a busy week in data flow, apparently. Evaluation (internal link) is uneventful; note that I reran a few performance outliers, which turned out not to be reproducible and hence were most likely wobbles.