Creation date in the codemeta.json generated by the tool https://autocodemeta.linkeddata.es/

[priya-gitTest]

Input : https://autocodemeta.linkeddata.es/
Output :

    "@context": "https://doi.org/10.5063/schema/codemeta-2.0",
    "applicationCategory": "SoftwareSourceCode",
    "codeRepository": "https://github.com/oeg-upm/auto-codemeta-generator",
    "dateCreated": "2025-03-05",
    "dateModified": "2025-04-04",
    "description": "This repository contains a (client-side) web application to automatically generate  Codemeta documents (aka. codemeta.json). This application is designed to work on popular modern browsers (Firefox,\nChromium/Google Chrome, Edge, Safari). Check [Caniuse](https://caniuse.com/)\nfor availability of features for these browsers. \n This repository is an extension of the Codemeta Generator to fill in the form automatically from a given GitHub or GitLab repository. \nThis repository contains a (client-side) web application to generate\nCodeMeta documents (aka. `codemeta.json`). \nThe [CodeMeta initiative](https://github.com/codemeta/codemeta) is a Free and Open Source academic collaboration\ncreating a minimal metadata schema for research software and code. \nThe academic community recommends on adding a codemeta.json file in\nthe root directory of your repository. \nWith this linked data metadata file, you can easily declare the authorship,\ninclude contextual information and link to other research outputs (publications,\ndata, etc.). \nAlso, the `codemeta.json` file in your source code is indexed in the\nSoftware Heritage (SWH) archive, which will improve findability in searches.\n \n",
    "downloadUrl": "https://github.com/oeg-upm/auto-codemeta-generator/releases",
    "keywords": [
        "codemeta",
        "metadata",
        "metadata-extraction",
        "research-software",
        "software"
    ],
    "license": "https://spdx.org/licenses/AGPL-3.0",
    "name": "auto-codemeta-generator",
    "programmingLanguage": [
        "JavaScript",
        "HTML",
        "CSS",
        "Python"
    ],
    "issueTracker": "https://github.com/oeg-upm/auto-codemeta-generator/issues",
    "@type": "SoftwareSourceCode"
}

Question : "dateCreated": "2025-03-05", Is this the date of creation of the fork ?
I also tried with the original Repo as input : https://github.com/codemeta/codemeta-generator
got the following :

   "@context": "https://doi.org/10.5063/schema/codemeta-2.0",
    "applicationCategory": "SoftwareSourceCode",
    "codeRepository": "https://github.com/codemeta/codemeta-generator",
    "dateCreated": "2020-02-28",
    "dateModified": "2025-03-28",

This of course does not match the codemeta.json file already present in both the repository and they have this date :

{
    "@context": "https://doi.org/10.5063/schema/codemeta-2.0",
    "@type": "SoftwareSourceCode",
    "license": "https://spdx.org/licenses/AGPL-3.0-only",
    "codeRepository": "https://github.com/codemeta/codemeta-generator",
    "dateCreated": "2019-10-02",

This date is definitely right wrt the original repo and its commit history.

So just curios what is the date we are targetting here? when fist commit is made or when a first release is created ? thanks

[dgarijo]

@priya-gitTest thanks for opening the issue. In this case, we return what the GitHub API returns.

In https://api.github.com/repos/codemeta/codemeta-generator you'll see the creation date being Mar 2020 instead of Oct 2019: "created_at": "2020-02-28T20:01:03Z"
I honestly don't have a proper explanation of why the creation differs... maybe we should elevate it to GitHub?

In auto-codemeta the assumption is that the codemeta file does not exist, hence we propose one. We are in the process of finding existing codemeta files to maybe extend them with missing stuff.

[dgarijo]

Interestingly, the license used is also different!
One is https://spdx.org/licenses/AGPL-3.0-only but the one returned by the GitHub API is https://spdx.org/licenses/AGPL-3.0

[priya-gitTest]

Thanks @dgarijo . I will check for the difference too and let you know if I find something. Need to understand what that date returned by API is.
I am assuming original codemeta.json is manually created now.
My question still remains, what is the right filed for this date from specification perspective. Is it the date when repository is created or when a first release is created . This also gets complicated when it is a Fork-ed repository.
May be it needs some revision ..https://codemeta.github.io/terms/

[dgarijo]

We use the GitHub codemeta crosswalk as the authoritative source: https://github.com/codemeta/codemeta/blob/master/crosswalks/GitHub.csv
I tried to find the source definition in https://docs.github.com/en/rest/repos/repos?apiVersion=2022-11-28, unsuccessfully.
My interpretation is the following:

• For a repo: created at represents the creation date
• For a release: created at represents the release date
• For a Forked repo: created at represents the date the repo was forked (i.e. created)
  Right now the problem is that GitHub gives a different date than the one from the first commmit.

[priya-gitTest]

I used some AI to check this and seems this [screenshot] is the only reliable option , if this is the date we wanted and I am not sure if their is an automated (API) way to get this .

I tried this too, but it didnt yielded the right date. GET https://api.github.com/repos/codemeta/codemeta-generator/commits?per_page=1&page=1&sort=created&direction=asc

[priya-gitTest]

But may be this repository has some complicated history. I will check later for one of my simple repo's and that should be easy to verify :)

[dgarijo]

Uf, relying on dumps in github is not the best way of doing this. All right, I will wait for your follow up.

[priya-gitTest]

1. For a simple repo, the API works fine !!! and date of creation via API = date of 1st/ Initial commit.
2. For a simple forked repo, with no commits, Date of Creation via API = Date of Forking the repository.

[dgarijo]

Our assumptions are correct then, and the crosswalk is fine. However, there may be issues in the GitHub API. We'll need to capture codemeta files to check for this in the future, since the codemeta field may be more authoritative than the one from the GitHub API (or not, who knows).

[priya-gitTest]

Well, I am checking with this Repo and cannot yet figure out, if there was a branch which got deleted or merge and it shows similar behavior as the 1st comment. so the Date of Creation via API != date of 1st/ Initial commit.
Ref : https://github.com/google/A2A

API returns : Created At: 2025-03-25T18:44:21Z
Main branch initial commit is on Apr 9, 2025

[priya-gitTest]

Our assumptions are correct then, and the crosswalk is fine. However, there may be issues in the GitHub API. We'll need to capture codemeta files to check for this in the future, since the codemeta field may be more authoritative than the one from the GitHub API (or not, who knows).

yes, perhaps need to verify this behavior across a few different vendors to get it right maybe.