I have a Sentry minidump of a crash from a user (truncated raw stack trace below). Using swift-5.9-RELEASE
A little complicated, but it's just an Electron app that loads a native module built from Swift. Don't worry, I'm packing the swift runtime dlls + vc redist + using a SxS manifest in the module I'm loading. I guess I can't totally ensure the user doesn't have Swift installed locally, but I'd really hope the SxS manifest would always prefer the specified side-by-side DLLs. The crash happens when we begin loading the native module, and the crash seems to not even touch our native Swift code it seems.
Thousands of users use this app daily without any issue, but this one user seems to be having issues. There are some locale-specific things going on here regarding the user (Chinese characters in paths, etc.) but we have other users in China successfully using the app, so that may be a red herring.
My question is, how do I symbolicate this? I have access to the suspect Swift runtime dlls in question, but when I try to upload them to Sentry, the sentry-cli doesn't detect any debug info. Just curious if anyone has had success using Sentry to debug the Windows Swift Runtime. I could simply bump the Swift runtime to a newer version and see if that does anything, but it's much better to not grasp at straws and just get to the bottom of what's going on.
Unfortunately, we do not have PDBs for the runtime yet. The problem here is that we need to figure out how to get those onto a symbol server. I do not wish to include them in the (offline) installer as that will grow the size of the installer - the installer is close to ~500 MiB now, and adding debug information would push it to ~20 GiB. This has a massive impact on experience, resource usage, and CI as well as I have been unable to convince GH to really include Swift as part of the default image.
That said, it should be able to symbolicate for anything in the export symbol list as that should still be present.