#I have a pipeline that started failing

1 messages · Page 1 of 1 (latest)

south folio
#

Does that come from the runtime or is it your own pipeline?

#

Oh, I see.

#

Try adding this to pyproject.toml (even though you're not using uv in there):

[tool.uv.sources]
dagger-io = { path = "sdk" }
lean hornet
#

thanks @south folio ill try that out

south folio
#

Would have figured this would have been caught by some of the tests that use requirements.lock but they run against a dev build and this should only be happening on prod build.

lean hornet
#

@south folio I just looked and this is what was already in there

[tool.uv.sources]
dagger-io = { path = "sdk", editable = true }

Should I omit editable, or is there some other issue?

south folio
#

Do you have both a requirements.lock and a uv.lock?

lean hornet
#

no just uv.lock

south folio
#

Then why is your pipeline doing process "uv pip install -e ./sdk -e . --no-deps -r requirements.lock?

lean hornet
#

My pipeline is not doing that I think its happening internally

#

This module is being called by another module

south folio
#

Ok, can you pinpoint which module is using the requirements.lock?

lean hornet
#

I dont have requirements.lock in any of my code

#

Ah wait, perhaps its a 3rd module

south folio
#

Yeah, it's what I mean. A dependency. miniflux?

lean hornet
#

Yeah there is a third module that has a requirements.lock

#

plausible

#

But possibly others too - there are like 12 modules being used here

south folio
#

I think the missing [tool.uv.source] is wiping the path to vendor.

lean hornet
south folio
#

No, it's a bug. I missed this.

lean hornet
#

Got it, let me try this and make sure it works. Ill report back!

south folio
#

We could do a hotfix release soon.