[🚀 Feature]: Selenium Manager support more architectures

[titusfortner]

Feature and motivation

Quick check of what architectures are supported by each driver...
Which of these should we support?
Do we need to wait until we can get rust and/or bazel to cross compile?
It looks like cross-compilation with cross might be straightforward - https://kerkour.com/rust-cross-compilation

Architecture	chromedriver	edgedriver	geckodriver
mac64	✅	✅	✅ 10%
linux64	✅	✅	✅ 48%
win32	✅	✅	✅ 5%
mac-arm64	✅	✅	✅ 10%
linux-arm64	❌	❌	🔲 2%
linux-32	❌	❌	🔲 1%
win-arm64	❌	✅	✅ 2%
win64	❌	✅	✅ 28%

[github-actions]

@titusfortner, thank you for creating this issue. We will troubleshoot it as soon as we can.

---

Info for maintainers

Triage this issue by using labels.

If information is missing, add a helpful comment and then I-issue-template label.

If the issue is a question, add the I-question label.

If the issue is valid but there is no time to troubleshoot it, consider adding the help wanted label.

If the issue requires changes or fixes from an external project (e.g., ChromeDriver, GeckoDriver, MSEdgeDriver, W3C), add the applicable G-* label, and it will provide the correct link and auto-close the issue.

After troubleshooting the issue, please add the R-awaiting answer label.

Thank you!

[titusfortner]

Well, played around a bit, and I could do:
https://github.com/titusfortner/selenium/actions/runs/3609798451

Note this builds x86_64-pc-windows-gnu but not x86_64-pc-windows-msvc

per the cross repo:

Additional Dockerfiles for other targets can be found in cross-toolchains. These include MSVC and Apple Darwin targets, which we cannot ship pre-built images of.

I could not figure out how to get those working, though; someone who knows more about docker than how to spell it might get further.

[ollie-iterators]

Is this related to the process of consolidating Selenium and Appium by making sure that Selenium can support the same architecture as Appium?

[titusfortner]

No, this is just making sure that people using Selenium on these architectures are automatically obtaining the correct drivers if publicly available.

[titusfortner]

@bonigarcia so this is essentially a rust question. Which of the drivers that get published is Selenium Manager looking for based on architecture? (e.g. does it always get win32 driver instead of the win64 geckodriver or edgedriver?)

I'm trying to figure out the different scenarios and what kinds of errors.

What driver does S/M return when:

1. Chrome on win-arm64
2. Edge on win-arm64

Are there combinations where nothing is obtained and an error is thrown? What is that error?
Are there unit tests for these scenarios?

[p0deje]

I'm playing around with Windows ARM and prebuilt SM downloads win32 for Chrome and win64 for Edge/Firefox:

PS C:\selenium> .\common\manager\windows\selenium-manager.exe --browser chrome --debug
DEBUG   Using shell command to find out chrome version
DEBUG   Running command: "wmic datafile where name='%PROGRAMFILES:\\=\\\\%\\\\Google\\\\Chrome\\\\Application\\\\chrome.exe' get Version /value"
DEBUG   Output: "\r\r\n\r\r\nVersion=112.0.5615.138\r\r\n\r\r\n\r\r\n\r"
DEBUG   The version of chrome is 112.0.5615.138
DEBUG   Detected browser: chrome 112
DEBUG   Required driver: chromedriver 112.0.5615.49
DEBUG   Running command: "chromedriver --version"
DEBUG   Output: "ChromeDriver 108.0.5359.71 (1e0e3868ee06e91ad636a874420e3ca3ae3756ac-refs/branch-heads/5359@{#1016})"
DEBUG   Running command: "where chromedriver"
DEBUG   Output: "C:\\ProgramData\\chocolatey\\bin\\chromedriver.exe"
WARN    Incompatible release of chromedriver (version 108.0.5359.71) detected in PATH: C:\ProgramData\chocolatey\bin\chromedriver.exe
DEBUG   chromedriver 112.0.5615.49 already in the cache
INFO    C:\Users\alex\.cache\selenium\chromedriver\win32\112.0.5615.49\chromedriver.exe

PS C:\selenium> .\common\manager\windows\selenium-manager.exe --browser firefox --debug
DEBUG   Using shell command to find out firefox version
DEBUG   Running command: "wmic datafile where name='%PROGRAMFILES:\\=\\\\%\\\\Mozilla Firefox\\\\firefox.exe' get Version /value"
DEBUG   Output: "\r\r\n\r\r\nVersion=111.0.1.8480\r\r\n\r\r\n\r\r\n\r"
DEBUG   The version of firefox is 111.0.1.8480
DEBUG   Detected browser: firefox 111
DEBUG   Required driver: geckodriver 0.33.0
DEBUG   Running command: "geckodriver --version"
DEBUG   Output: ""
DEBUG   Driver URL: https://github.com/mozilla/geckodriver/releases/download/v0.33.0/geckodriver-v0.33.0-win64.zip
DEBUG   File extracted to C:\Users\alex\.cache\selenium\geckodriver\win64\0.33.0\geckodriver.exe (3884960 bytes)
INFO    C:\Users\alex\.cache\selenium\geckodriver\win64\0.33.0\geckodriver.exe

PS C:\selenium> .\common\manager\windows\selenium-manager.exe --browser edge --debug
DEBUG   Using shell command to find out edge version
DEBUG   Running command: "wmic datafile where name='%PROGRAMFILES(X86):\\=\\\\%\\\\Microsoft\\\\Edge\\\\Application\\\\msedge.exe' get Version /value"
DEBUG   Output: "\r\r\n\r\r\nVersion=112.0.1722.48\r\r\n\r\r\n\r\r\n\r"
DEBUG   The version of edge is 112.0.1722.48
DEBUG   Detected browser: edge 112
DEBUG   Required driver: msedgedriver 112.0.1722.48
DEBUG   Running command: "msedgedriver --version"
DEBUG   Output: ""
DEBUG   Driver URL: https://msedgedriver.azureedge.net/112.0.1722.48/edgedriver_win64.zip
DEBUG   File extracted to C:\Users\alex\.cache\selenium\msedgedriver\win64\112.0.1722.48\msedgedriver.exe (17087912 bytes)
INFO    C:\Users\alex\.cache\selenium\msedgedriver\win64\112.0.1722.48\msedgedriver.exe

[titusfortner]

So we should be getting the win-arm64 versions of msedgedriver and geckodriver.
Should we be erroring for Chrome since there isn't a win-arm64 version? Or should we let the driver see what it can do?

[p0deje]

So we should be getting the win-arm64 versions of msedgedriver and geckodriver.

Yes, we should

• msedgedriver supports ARM64 https://developer.microsoft.com/en-us/microsoft-edge/tools/webdriver/
• GeckoDriver supports win-aarch64 https://github.com/mozilla/geckodriver/releases/tag/v0.33.0

Should we be erroring for Chrome since there isn't a win-arm64 version? Or should we let the driver see what it can do?

I guess the current behavior is fine as long ChromeDriver works.

[titusfortner]

Do we know if it does? I'm kind of curious why Google doesn't release a separate binary, or what might not work. I guess all we can do is give the best option available and figure it out from there.

[p0deje]

I suppose it works given Ruby tests for Chrome pass on my Windows VM.

[bonigarcia]

Regarding chromedriver, it is ok that Selenium Manager in Windows always downloads the x32 architecture since that's the only one published by the chromedriver team.

Regarding geckodriver and msedgedriver in Windows, Selenium Manager in Windows currently only supports x32 and x64 architectures. I can prepare a patch to support ARM64 for those Windows drivers (although I don't have an ARM64 machine to test with).

[bonigarcia]

@p0deje Can you run the following command in your Windows ARM64 and tell me the output, please?

wmic os get osarchitecture

[p0deje]

@bonigarcia There you go, I can test this as soon as you have a new version (I can also build manager now)!

C:\selenium>wmic os get osarchitecture
OSArchitecture
ARM 64-bit Processor

[bonigarcia]

Thanks @p0deje.

I have just committed a patch to enable the detection of ARM in Windows (branch se_mgr_winarm). You can get the resulting binary using that branch from here:

https://github.com/SeleniumHQ/selenium/actions/runs/4787671731

[titusfortner]

I updated the original post. Green checkmarks are the ones I'm pretty sure that Selenium Manager downloads right now. Red X's aren't supported by the driver. The empty boxes are the ones that have a driver, but I don't know if SM will grab them.

@bonigarcia can you verify which drivers can be downloaded right now based on architectures we are parsing? Thanks!

[p0deje]

@bonigarcia It works great!

C:\Users\alex\Downloads>selenium-manager.exe --browser edge   
INFO    C:\Users\alex\.cache\selenium\msedgedriver\win-arm64\112.0.1722.58\msedgedriver.exe

C:\Users\alex\Downloads>selenium-manager.exe --browser firefox
INFO    C:\Users\alex\.cache\selenium\geckodriver\win-arm64\0.33.0\geckodriver.exe

[bonigarcia]

@titusfortner The empty boxes on win64 should be green. Also, with the latest patch (branch se_mgr_winarm), the empty boxes on win-arm64 should be green as well.

[titusfortner]

So it's just the geckodriver, linux 32 & Linux ARM that aren't getting used?

[bonigarcia]

So it's just the geckodriver, linux 32 & Linux ARM that aren't getting used?

That's right.

[titusfortner]

Are those easy to toggle for/add? Would love to close this issue and not think/worry about architectures any more. 😂

[bonigarcia]

Are those easy to toggle for/add?

I think these drivers would be supported if these Selenium Manager binaries (i.e. Linux 32 and Linux ARM64) would be generated.

[titusfortner]

Can we cross compile Linux 32 bit on the 64bit container? I think(?) it might work for all Linux implementations similar to what we do for windows?

[bonigarcia]

Can we cross compile Linux 32 bit on the 64bit container?

I didn't try, but I suppose it is possible.

I think(?) it might work for all Linux implementations similar to what we do for windows?

I'm not sure about that, perhaps it is possible, like in Windows, but I didn't try.

[bonigarcia]

Do we plan to support any other architecture (e.g. Linux ARM) at this moment? If not, can we close this issue?

[titusfortner]

Does win-arm64 work for Edge & Firefox?

The ones that aren't supported are Linux 32 bit & Linux ARM, and yeah, I think we're ok ignoring those for now.

[bonigarcia]

Does win-arm64 work for Edge & Firefox?

I don't understand this question

[titusfortner]

@p0deje did you say that the windows binary we have works for you on your M1 in Windows VM?
(and does it properly get the arm versions of Edge & Firefox drivers?)

[p0deje]

@titusfortner yes, it works fine #11357 (comment)

[titusfortner]

So, yeah. we support 100% of chrome & edge drivers & 97% of geckodriver downloads. We can close.
🎉 🚀

[github-actions]

This issue has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.