WIP: Change namedtypes #3147
WIP: Change namedtypes #3147
Conversation
There was a problem hiding this comment.
Hello, and thank you for opening this PR! 🎉
All contributors have signed the CLA, thank you! ❤️
Commit Messages
We want to keep history, but for that to actually be useful we have
some rules on how to format our commit messages (relevant xkcd).
Please stick to these guidelines for commit messages:
- Separate subject from body with a blank line
- When fixing an issue, start your commit message with
Fix #<ISSUE-NBR>:- Limit the subject line to 72 characters
- Capitalize the subject line
- Do not end the subject line with a period
- Use the imperative mood in the subject line ("Added" instead of "Add")
- Wrap the body at 80 characters
- Use the body to explain what and why vs. how
adapted from https://chris.beams.io/posts/git-commit
Have an awesome day! ☀️
The problem is that the assertion in checkNoLeaks might be wrong if the widened type is erroneous. This happened during editing.
Also - rename Fresh flag to NonMember - update criterion for phases where refs are symbolic. This used to be just before erasure (at resolveSuper) but then other phases were inserted after resolveSuper and the test became outdated. It's more robust now.
odersky
force-pushed
the
change-namedtypes
branch
from
7cd9e3a to
8d8491a
Compare
odersky
force-pushed
the
change-namedtypes
branch
from
af4dc3e to
2cee5a7
Compare
- simplify Substituters - use sym.isOverridable as a test instead of checking whether the reference is a WithFixedSym.
odersky
force-pushed
the
change-namedtypes
branch
from
6bedb01 to
8971f1f
Compare
|
This PR turned out to be a dead-end, and the approach in #3087 and #3134 is preferable. The issue is that, as long as we base our tooling on Tasty we cannot define references by their symbols, since these references can go across compilation units. So we need the full distinction between references as names, names with signatures and/or namespaces, and symbols. There's no way around it. We just have to make it as streamlined as we can. |
This is a second attempt to simplify named types. The first attempt ran into lots of complications when trying to accommodate private members. I decided it's better to start from scratch.
The new attempt tries to keep the basic layout of a NamedType as a combination of prefix and name, without fiddling with designators. Instead we want to base more functionality on the symbol in the type.