Hello team 👋 I have a

python_distribution

that includes a large number of

resource

targets - about 31,000 YAML files. This package takes a staggering 80 minutes to run

package

. If I remove the

resource

targets, then

package

runs instantly. 31,000 YAML files is a lot of files, but since these are all

resource

targets,

pants

should not actually need to read any of them, it should only need to list them (

pants list

on these targets runs instantly), add them to the

.whl

and list them in the manifest. What is it doing that takes so long?

Do they end up in the sandbox? My guess off the cuff would have been materializing them into the sandbox

The 31,000 YAML files are included in the final

.whl

so I have no doubt that they do Confirmed yes, the 31,000 YAML files are copied to the sandbox, but it should not take very long to copy these files. There's a lot of files, but each one is very small. The final

.whl

is 31MB

Can you share a redacted version of this repo so we can benchmark on it directly?