SPMI: Extend list of unrecorded environment variables #77504
Conversation
We are currently hitting issues when there are diffs in the FileCheck tests because they are recorded with JitStdOutFile and the SPMI driver does not override this correctly. Add this and a lot of other environment variables to the list of variables we do not record with. This big list of string compares may look somewhat scary but we only record variables once during startup of the JIT, so it shouldn't be too bad.
dotnet-issue-labeler
bot
added
the
area-CodeGen-coreclr
ghost
assigned 
|
Tagging subscribers to this area: @JulieLeeMSFT, @jakobbotsch Issue DetailsWe are currently hitting issues when there are diffs in the FileCheck tests because they are recorded with JitStdOutFile and the SPMI driver does not override this correctly. Add this and a lot of other environment variables to the list of variables we do not record. This big list of string compares may look somewhat scary but we only query the config once during startup of the JIT, so it isn't a problem. I was not able to measure anything significant when collecting a simple "Hello, world!" C# app. For this list I simply went through the variables and added the ones I don't think it makes sense to record into the collection. Let me know if any of them seem unreasonable.
|
| // Special cases: we don't want to store some DOTNET variables during | ||
| // collections, typically when they refer to file paths or simply because | ||
| // it does not make sense to replay with it. | ||
|
|
There was a problem hiding this comment.
I verified that DOTNET_JitName and DOTNET_AltJitName do not come through this path since the JIT doesn't query them. So they don't need to be here.
We are currently hitting issues when there are diffs in the FileCheck tests because they are recorded with JitStdOutFile and the SPMI driver does not override this correctly. Add this and a lot of other environment variables to the list of variables we do not record.
This big list of string compares may look somewhat scary but we only query the config once during startup of the JIT, so it isn't a problem. I was not able to measure anything significant when collecting a simple "Hello, world!" C# app.
For this list I simply went through the variables and added the ones I don't think it makes sense to record into the collection. Let me know if any of them seem unreasonable.