25 Apr 2024 |
diemol | Seems a unit test is failing | 22:03:58 |
26 Apr 2024 |
simonstewart | I chew away on how to do this in the java code sometimes. We have a bunch of async code, but we’d like to raise exceptions on the main thread. Nicely done cracking that nut in .Net! 🙂 | 01:17:47 |
simonstewart | https://github.com/SeleniumHQ/selenium/pull/13879 | 01:47:34 |
simonstewart | Until we migrate the atoms to TypeScript, I think we’re stuck with google closure | 01:47:57 |
p0deje | Can we avoid rules_closure_dependencies altogether? | 01:48:54 |
p0deje | Ah, it’s Closure, not Proto, nevermind then! | 01:49:19 |
p0deje | I wonder how hard is it to convert from Closure to TS using https://github.com/angular/clutz?tab=readme-ov-file#gents---closure-to-typescript-converter | 01:50:45 |
simonstewart | I think AutomatedTester tried it before without success | 01:53:11 |
p0deje | I’ve released Ruby bindings 4.20.1 with a fix https://github.com/SeleniumHQ/selenium/issues/13876 | 01:54:55 |
diemol | But this
ENV['CHROMEDRIVER_BIN'] = Selenium::WebDriver::DriverFinder.path(options, Selenium::WebDriver::Chrome::Service).driver_path
Is still failing | 05:26:36 |
diemol | The static method is back but the invocation produces the same error | 05:27:31 |
AutomatedTester | I didn’t get far because I was put off by the aspect dev stuff.
There is a lot of dead code in the atoms that if we cleaned up would make the migration simpler | 06:28:01 |
simonstewart | The aspect migration has now been done. | 06:33:16 |
simonstewart | I shall share the horror stories once we’re all in the same place again | 06:33:45 |
AutomatedTester | Closure is still screwing the pooch for me...
I have the latest HEAD and have done bazel clean --expunge and then bazel test //py:unit and still get the linker error | 10:13:48 |
simonstewart | Just having dinner. Once I’m done, want to pair on it? | 10:15:41 |
AutomatedTester | yep sounds good but also... I don't want to take you from your family | 10:16:09 |
simonstewart | My work day is split into two bits anyway so I have some overlap with my team back home. | 10:16:44 |
jimevans | Credit where it's due, I did take some inspiration from the Puppeteer .NET implementation. If people are interested, I'm happy to discuss it at the summit. | 13:05:37 |
p0deje | You shouldn’t call .driver_path in the end | 13:45:28 |
p0deje | That wasn’t part of original code, was it? | 13:46:12 |
diemol | Ah ok | 13:47:21 |
diemol | you are right | 13:47:23 |
diemol | How does the new invocation look like? | 13:47:42 |
diemol | Like this?
finder = Selenium::WebDriver::DriverFinder.new(options, Selenium::WebDriver::Chrome::Service.new) | 13:48:49 |
diemol | Yes, that works | 13:49:44 |
diemol | I need to learn how to read code | 13:49:52 |
diemol | Conceptually, why is that different?
https://github.com/SeleniumHQ/seleniumhq.github.io/commit/9b3fc3606afb0d379b5d4d8cbc405d8fa9ddc60c | 14:19:01 |
diemol | Why I needed the new in the service code I had before your change. | 14:19:22 |
diemol | Is the new implicit when this is done?
service = Selenium::WebDriver::Service.chrome | 14:19:51 |