Test on Linux with Swift 5.3, bump SwiftSyntax

NSHipster / SwiftSyntaxHighlighter

A syntax highlighter for Swift code that uses SwiftSyntax to generate Pygments-compatible HTML.

https://nshipster.com/swiftsyntax/

MIT License

275 stars 12 forks source link

Test on Linux with Swift 5.3, bump SwiftSyntax #8

Closed MaxDesiatov closed 3 years ago

mattt commented 4 years ago

@MaxDesiatov Nice! Just waiting for the ✅ from CI. (So nice to finally have an official SwiftSyntax tag for 5.3)

mattt commented 4 years ago

@MaxDesiatov Thanks for trying to get backwards compatibility to work. For this dependency, I'd be very happy to drop support for 5.2 with a new major / minor version.

MaxDesiatov commented 4 years ago

That crash on Linux is quite puzzling. At first I though that it's related to the lack of the libxml package that swift-doc CI config installs, but that doesn't seem to make any difference. 🤔

mattt commented 4 years ago

@MaxDesiatov Yeah, I was confused by that too. "Exited with signal code 4" makes me think that it's an assertion failure for a failing force unwrap. But I don't know why we aren't getting a backtrace in the logs...

Edit: Wait, I think I know why. https://docs.github.com/en/actions/managing-workflow-runs/enabling-debug-logging

Edit 2: Or yeah, maybe it's because we're testing in release mode. IIRC, I had it set up that way to avoid any dev/prod issues with the distributed executable.

MaxDesiatov commented 4 years ago

Testing with debug builds didn't help ¯\_(ツ)_/¯

Anyway, I'm off for today, I hope I'll be able to have a closer look in the next few days. Thanks for reviewing these PRs!

mattt commented 4 years ago

@MaxDesiatov Same here. Thanks again for all of your help! It's been fun hacking on this together.

mattt commented 3 years ago

@MaxDesiatov According to GitHub, "Secrets are not passed to workflows that are triggered by a pull request from a fork." [1], so I don't think ACTIONS_STEP_DEBUG is being set. I'm retargeting this PR to a new swift-5.3 branch in this repo so that we can get some visibility into what's causing CI to fail on Linux.