Markus Stange

Ok, let's look at these four features one by one. The four features being: 1. The ability to load source from a public server on the web (github) for Java/Kotlin...

Oh I see, 3 has nothing to do with source maps. Ok. So the API call is more about "I'm notifying you that the user is currently looking at a...

Sure, then let's make two different API calls for two different use cases: One API call for obtaining source code and a different API call to navigate the IDE. We...

Oh, Jeff noticed a bug when testing the deploy preview: When a preview selection is applied, the percentages are currently of the non-preview-filtered thread, which is inconsistent with the call...

> My opinion is that we should just always display the category tooltip and completely remove anything related to the implementation, but we can do this independently of this PR....

Nice! I have some suggestions for the points you raised in the initial comment: it basically comes down to drilling backend-specific holes in the API. > as a workaround we...

> provided someone is willing to go over and write all the boilerplate Here's a partial attempt I made a few weeks ago: https://github.com/mstange/zstd-rs/commits/displaydoc-instead-of-derive_more/ It uses `displaydoc::Display` and manually implements...

Looking at the history, it was actually `thiserror-core` that you switched away from, not `thiserror-no-std`. The former requires Nightly for `no_std` builds, the latter just doesn't implement any `Error` trait...

@jan-auer Could you take a look at this? It's needed for https://github.com/mstange/pdb-addr2line/pull/63

The first thing I'd try here would be to update the dump_syms dependency. dump_syms 2.0 uses a newer symbolic-debuginfo which uses a newer pdb, which might have replaced this panic...