SHA hashes in 'main' branch are different from those in the 'origin/main'

[KarelVesely84]

Hello,
i am experimenting with Espresso, trying the switchboard recipe on air-traffic-control data.
I noticed my local SHA hashes in 'main' branch are different from those in the 'origin/main'.
I tried to 'pull' from the 'origin', but i am getting conflicts due to that.

• Can it be caused by the local installation via pip install --editable . ?
• Have you seen this issue before ? Is it normal, or did I do something wrong ?
• How do you edit code, contribute and test locally normally ?

Best regards,
Karel

[freewym]

Hi Karel,

I periodically sync this repo with fairseq's official repo with "rebase" so that all my commits are always on top of the official fairseq ones. Th purpose is to keep this repo up-to-date while making it easy to track the changes in Espresso. However, I understand it will make it hard for users to sync with this repo.

Currently there is only me contributing to this repo. I usually test it manually (by running a small recipe locally)

[KarelVesely84]

Hello Yiming,
thank you for exaplaining the situation.

Aha, after you rebased, i fetched the origin/main,
looked up the SHA of the last common ancestor
and rebase to that. This worked fine...

Other option would be to use fairseq as a git submodule
(but I see that fairseq was modified in your commits on top of fairseq).

The nice thing about git submodules is that they are fixed to a specific
commit by SHA code, and the updates are done manually. So the
versioning and updating is 'under control'
(I created a toy example to get some idea how it works, actually)

Thank you, best regards,
Karel