Executed "map.slt". Took 460.5636141s
Executed "window.slt". Took 217.240136s
Executed "struct.slt". Took 1316.049712s
Executed "array.slt". Took 2407.7054095s

Thats really weird, a simple array SLT takes 40 mins

@comphead disregarding build arguments, which may affect build performance, there seems to be a known issue with IO on windows runners which is noticeably slower than linux/mac ones. From comments to that issue it looks like some projects are able to get performance benefit using DevDrive, so it may be a different perspective to look at this problem from.

@comphead disregarding build arguments, which may affect build performance, there seems to be a known issue with IO on windows runners which is noticeably slower than linux/mac ones. From comments to that issue it looks like some projects are able to get performance benefit using DevDrive, so it may be a different perspective to look at this problem from.

Thanks @korowa for the hint I was playing with lots of combination including different Windows images but nothing seems to work.

Some projects give up rust testing on Windows like DataDog/orchestrion#415

@alamb wondering should we pause the Windows flow temporarily? The build time is incredibly slow and we holding GH resources for way longer time

Another thing that would benefit the project as a whole is to reduce build time.

One thing I noticed a while ago is that each binary built (e.g. every example) was around 100MB

I was thinking that we could reduce / conslidate the number of examples / binaries built which would speed up all runners (not just Windows)

For example consolidating https://github.com/apache/datafusion/blob/main/datafusion-examples/examples/regexp.rs and other examples of using functions into dataframe_api.rs might be a good win

Closing it as there is no any progress after all the experiments