Aria Desires

In the current impl if we get two files we only use the "better" one (determined by the caller's ordering of the inputs). In this impl we pass the debugfile...

TODO: * [ ] clean up all the nonsense around trying to make the cache/tmp optional. this feels like a failed experiment in the current form, but if we change...

I am currently in the "if it ain't broke..." camp on this. But if we ever write our own rust-based macos minidump writer, this might be painless enough to be...

Notes: main rangemap construction code here, which does some special pruning and fixups: https://github.com/rust-minidump/rust-minidump/blob/e0f8b5a7b4b6d9eebdf297f59c7aaa49c93f8494/minidump-common/src/traits.rs#L68-L112 breakpad-symbols is soft-forking some of the code here: https://github.com/rust-minidump/rust-minidump/blob/e0f8b5a7b4b6d9eebdf297f59c7aaa49c93f8494/breakpad-symbols/src/sym_file/parser.rs#L571-L588 and also does some manual hacky corrections...

That won't correctly handle the truncating case in the last file I linked, which is the major one and a real problem in production.

My condolences on the rangemap stuff. I too spent a while being like "why the fuck is this like this, I should just rip this all out... ugh no there's...

#289 got everything piped and wired up to the right place, but for now we're punting on actually parsing out the extra sections, absent a reason to care about those...

Since this could change the way we expose contexts, it's a 1.0 blocker

Other useful things for this: The printing code shows you how to enumerate/analyze the xstate entries mechanically: https://github.com/luser/rust-minidump/blob/16a7f1b392ed998ad0d02054f27770f29e066a56/minidump/src/minidump.rs#L3164-L3176 And this comment details how to interpret the xstate entries *logically* (they're...

this interface design is also desirable for shipping #263