Temporarily suppress dependent name types #329
Conversation
|
|
||
| def test_issue328(self): | ||
| self._test_issue328(py11) | ||
| self._test_issue328(boost) |
There was a problem hiding this comment.
In situations where we are just suppressing an issue instead of solving it, could we add an assertion that should work if we didn't have the error (like being able to access this ::type) and then comment it out in the test?
I am worried it will be hard to tell which issues we have solved and which ones we have just covered up. This makes it hard if we e.g. upgrade clang: some of this functionality might get fixed, but we won't know how to test if it is without reading many old bug reports.
I fully agree with you. Sorry, I should have provided more details. The intention is actually resolving the issue in the end instead of just concealing it. Supporting the built-in type (#327) is in the path for example. This PR is just to suppress throwing the exception for now. I think we should print a warning saying that the API is ignored as it's not supported yet in the meantime. At a high level, I would like to extend the chimera features to reasonably support C++ projects with lots of templates. As an example, I tried to use chimera for FCL 0.6, and this segfault the first error I got. 😞 |
jslee02
changed the title
|
Ok, that makes sense. Perhaps just add that placeholder unit test that shows the problem as per https://github.com/personalrobotics/chimera/pull/329/files#r419865135 and it should be fine to merge. |
Related issue: #328