Loose boto3 requirements #457
Conversation
|
Does it solve the problem with aws-cli? |
|
@PawelPeczek-Roboflow yes! You can see here it uses the first version found. https://github.com/roboflow/inference/actions/runs/9416030487/job/25938277946#step:4:136 |
|
amazing |
| @@ -23,6 +23,6 @@ pydantic~=2.6 | |||
| openai>=1.12.0 | |||
| structlog>=24.1.0 | |||
| zxing-cpp>=2.2.0 | |||
| boto3<=1.28.23 | |||
There was a problem hiding this comment.
could we align with current max version such that we do not move so far into the unknown versions
There was a problem hiding this comment.
Sure! I limited to current version!
Now that we understand the problem, we just have to remember to sometimes bump this limit. Or we could limit botocore and awscli versions too, because they're implicitly limited anyway.
There was a problem hiding this comment.
I am fine with both, yet prefer explicit
|
U are free to merge once those run without issues: |
|
all 🟢 - u are free to merge |
Description
Bump the max boto3 version to avoid pip trying to install many versions before finding one that works.
It is reasonably safe, boto3 is very stable.
Type of change
How has this change been tested, please provide a testcase or example of how you tested the change?
🟢 CI pass 🟢