Is it possible to remove broken label from labels list?

[StrikerRUS]

Comment:

Due to the old typo in the version field (fixed in #49) there is a broken label at the package page.

This results in a wrong version badge for the package:

It can be solved via specifying label=main in shields.io badge constructor, but this feature isn't released yet: badges/shields#2955.

[jameslamb]

Others can correct me if I'm wrong... but I think that label needs to stay, to prevent conda install from finding the package.

From the conda-forge docs:

As a matter of policy, we do not allow edits or the deletion of packages on conda-forge. This policy is very important as it increases the reliability and reproducibility of conda environments made with the conda-forge channel

If you need to remove a package, please see the section on marking packages broken.

I believe we're stuck with that package being there, and will need that change in badges/shields#2955 to get a conda-forge-specific badge with the correct label.

[StrikerRUS]

Looked at https://api.anaconda.org/package/conda-forge/lightgbm and noticed the following:

...
    "versions": [
        "2.0.4",
        "2.0.5",
        "2.0.7",
        "2.0.10",
        "2.1.0",
        "2.1.1",
        "2.1.2",
        "2.2.0",
        "2.2.1",
        "2.2.2",
        "2.2.3",
        "2.3.0",
        "2.3.1",
        "3.0.0",
        "3.1.0",
        "3.1.1",
        "3.2.0",
        "3.2.1",
        "3.3.1",
        "3.3.2",
        "3.3.3",
        "3.3.4",
        "3.3.5",
        "4.0.0",
        "bee59dd269a93b093f2c610d4a6683a7ea87c63d3ea35c622123ce2c020b2abc",
        "4.1.0",
        "4.2.0",
        "4.3.0",
        "4.4.0"
    ],
    "latest_version": "bee59dd269a93b093f2c610d4a6683a7ea87c63d3ea35c622123ce2c020b2abc",
    "platforms": {
        "linux-64": "bee59dd269a93b093f2c610d4a6683a7ea87c63d3ea35c622123ce2c020b2abc",
        "linux-ppc64le": "bee59dd269a93b093f2c610d4a6683a7ea87c63d3ea35c622123ce2c020b2abc",
        "linux-aarch64": "bee59dd269a93b093f2c610d4a6683a7ea87c63d3ea35c622123ce2c020b2abc",
        "win-64": "bee59dd269a93b093f2c610d4a6683a7ea87c63d3ea35c622123ce2c020b2abc",
        "osx-64": "bee59dd269a93b093f2c610d4a6683a7ea87c63d3ea35c622123ce2c020b2abc",
        "osx-arm64": "bee59dd269a93b093f2c610d4a6683a7ea87c63d3ea35c622123ce2c020b2abc"
    },
...

I guess it's a bug. Probably latest_version field is computed by sorting versions alphabetically.

This field is used by badges/shields:
https://github.com/badges/shields/blob/81dfc92f86c7d71d369af50c84f675687b1093fe/services/conda/conda-version.service.js#L49

[StrikerRUS]

Found one similar issue about versions inconsistency: anaconda/anaconda-client#561. Unfortunately, there are no responses in that issue.

Does anyone know which script is responsible for creating those json-files from https://api.anaconda.org/package/? It seems there is a lot of room for improvement.

[StrikerRUS]

Looks like 4.5.0 release helped to place correct value in the latest_version field.

[jameslamb]

Probably latest_version field is computed by sorting versions alphabetically.

Yes I think so. I just commented something similar on the issue you linked: anaconda/anaconda-client#561 (comment)

Does anyone know which script is responsible for creating those json-files from https://api.anaconda.org/package/?

I don't .... BUT I did find an issue over in conda/infrastructure with active, recent participation from folks at Anaconda. I just summarized our case here in a comment on that issue: conda/infrastructure#943 (comment)

Since https://api.anaconda.org/package/conda-forge/lightgbm is reporting the correct latest_version now and lightgbm's badge is correct, I think we should close this issue and keep discussion over in conda/infrastructure#943.

So I'm going to close this, but can re-open it if you disagree @StrikerRUS . Thanks for your investigation and all the links with supporting evidence!

[StrikerRUS]

@jameslamb Thank you so much for finding the right place to report this problem!
Sorry, I couldn't do that.

[jameslamb]

No problem! You initial investigation was very helpful!