Change the default SuperPMI collection location#78513
Conversation
When using `superpmi.py collect`, change the default output MCH file to be located in the current directory, not the "spmi" location. Also, add logic so multiple concurrent collections using a default file name don't overwrite each other.
|
Tagging subscribers to this area: @JulieLeeMSFT, @jakobbotsch Issue DetailsWhen using
|
|
@jakobbotsch @dotnet/jit-contrib PTAL |
|
@dotnet/jit-contrib anyone? |
Could you confirm what is the motivation behind this change? Is it just the convivence or you hit any problems with it? |
| else: | ||
| # Default directory is the current working directory (before we've changed the directory using "TempDir") | ||
| default_mch_location = os.path.abspath(os.getcwd()) | ||
| if not os.path.isdir(default_mch_location): |
There was a problem hiding this comment.
In what case os.getcwd() would return a location that is not a directory? When it is ran from C:\ or something?
There was a problem hiding this comment.
Yeah, getcwd should already return an existing directory.
It's pure convenience. If I run "superpmi.py collect repro.bat" I want the resulting MCH file to be in the current directory, not in some directory near all the cached, downloaded MCH files. Of course, you can specify |
When using
superpmi.py collect, change the default output MCH file to be located in the current directory, not the "spmi" location. Also, add logic so multiple sequential collections using a default file name don't overwrite each other.