Boris Staletic

For figuring out the system headers, cquery does take into account `--gcc-toolchains`, `--sysroot` and `-nostdlib`. https://github.com/cquery-project/cquery/blob/master/src/project.cc#L123-L128

No, [cquery passes `results.args` as the `flags` for `GetSystemIncludes`](https://github.com/cquery-project/cquery/blob/master/src/project.cc#L439-L440). [The `results.args` is populated by cquery by sanitizing your flags](https://github.com/cquery-project/cquery/blob/master/src/project.cc#L345-L437). Then [cquery loops through the sanitized flags and cherry-picks what flags...

Shouldn't the "all drivers that support GLES3.0" be changed to explicitly list the drivers too? Otherwise mesamatrix reports red for all drivers.

Your PR landed, the field in the "mesa" column is green, but all drivers are red, even though some should be green.

Thanks for explaining what's going on. Like I said, this isn't a big deal for me. > Maybe it should be reported to Vim's issue tracker; I'm not sure how...

For the record, this can be fixed, while staying... ugh... bug compatible with vscode\*. [`JDTDelegateCommandHandler`](https://github.com/eclipse/eclipse.jdt.ls/blob/master/org.eclipse.jdt.ls.core/src/org/eclipse/jdt/ls/core/internal/JDTDelegateCommandHandler.java) is a catch-all command handler. It's pretty trivial to extend it to support `java.apply.workspaceEdit` ```diff...

As I mentioned in #1939, this regression happened with https://github.com/eclipse/eclipse.jdt.ls/pull/1445 That PR did introduce "if file doesn't exist, `return`". Hope this can be of some help.

Hi. I'm aware that the behaviour of `split-or-existing-window` isn't always perfect. Is what you're asking for basically "listen to `` and don't be a smartass"?

`mods` are what you called "prefix" in your first post. See `:h `.

For the record, I made this pull request well after midnight. I might have done something stupid.