docs: Migrate docs from xlnx-main

[pamolloy]

• Migrate docs from Xilinx development branch
• Document kernel versions that are available, upstream stable releases, how to pick a version, etc

[gastmaier]

We re-use pages from the kernel Documentation/, what do you suggest? drop this approach? or make the CI fancy (e.g., aggregate from multiple branches)?

[gastmaier]

On important thing is that sparce checkout does fetch all files at head, the only thing it does is to not put all at the current worktree.
So, if github default checkout I would say no to aggregate from multiple branches.
Still, we have the fancy cached-checkout to speed things up.

Or play with

https://api.github.com/repos/analogdevicesinc/linux/git/trees/main
# get sha for path 'Documentation', e.g.
https://api.github.com/repos/analogdevicesinc/linux/git/trees/99dd308027d28eb8e006f4fde949c9fbf4b28f19
# same for iio
https://api.github.com/repos/analogdevicesinc/linux/git/trees/6a15237a2487556e1e17002d2a984493d61167fc
# Yippie! enjoy your raw files! :)

And curl raw files

import requests

REPO = "analogdevicesinc/linux"
TARGET_PATH = "Documentation/iio"
BASE_URL = f"https://api.github.com/repos/{REPO}/git/trees"

def get_sha_for_path(tree_sha, path_segment):
    url = f"{BASE_URL}/{tree_sha}"
    response = requests.get(url).json()
    
    for item in response.get('tree', []):
        if item['path'] == path_segment:
            return item['sha']
    return None

current_sha = "main"

for segment in TARGET_PATH.split('/'):
    current_sha = get_sha_for_path(current_sha, segment)
    if not current_sha:
        print(f"Path segment '{segment}' not found.")
        exit()

final_tree_url = f"{BASE_URL}/{current_sha}"
files = requests.get(final_tree_url).json()

rst_files = [f for f in files['tree'] if f['path'].endswith('.rst')]

for f in rst_files:
    print(f"File: {f['path']} | SHA: {f['url']}")

then for a url

import requests
import base64

url = "https://api.github.com/repos/analogdevicesinc/linux/git/blobs/9a489a79d8f5a8682334266ab8c444eeddd51773"
response = requests.get(url).json()

file_content = base64.b64decode(response['content']).decode('utf-8')
print(file_content)

We can create a action like

    steps:
    - uses: analogdevicesinc/doctools/fetch-tree@action
        with:
          # what to fetch
          fetch: |
            Documentation/iio
            Documentation/other/path
          # write to
          path: /tmp/some/path

[nunojsa]

The commit hash in the commit seems to reference the wrong commit...

or make the CI fancy (e.g., aggregate from multiple branches)?

Is there a real need for that? I would say not to overcomplicate things unless we really have too.

[nunojsa]

Store ADI specific documentation for all product groups on the CI branch
rather than separately on each branch. This is also necessary for
products like ADSP that do not maintain a main development branch

Ok, I see now the point...

[gastmaier]

Then, can you delete all files that include ../../Documentation?

[gastmaier]

let's do the wipe first. and leave the 'aggregate' for a future pr

[pamolloy]

The commit hash in the commit seems to reference the wrong commit...

It is the latest commit on main?

https://github.com/analogdevicesinc/linux/commits/main/

[nunojsa]

It is the latest commit on main?

Ahaha, I get now what you mean in the message 🤦‍♂️. But you do leave some room for false interpretation 😄

[pamolloy]

Still plenty to do, but just pushed what I have so far about CVEs and stable versions. Found the last two quotes from the LWN article especially interesting.

2be4bee

[gastmaier]

Still plenty to do, but just pushed what I have so far about CVEs and stable versions. Found the last two quotes from the LWN article especially interesting.

2be4bee

Ok, but why not further prs adding more docs?