markhoerth commented on PR #12399:
URL: https://github.com/apache/gravitino/pull/12399#issuecomment-5246228668

   @yuqi1129 You were right, and I have restored the five per-cloud pages. 
Pushed just now.
   
   What I kept from the consolidation is the reference material. The property 
tables, the fileset model, schema and fileset properties, and property 
inheritance now live once in fileset-catalog.md rather than being copied five 
times. That was the drift problem I was trying to solve, and it is solved 
without touching the user journey.
   
   What is back is the task. Each per-cloud page carries the full runnable 
path: install the bundle, create the catalog, create the schema, create the 
fileset, then read it through the Java client, Spark, the Hadoop command line, 
and Python, with the credential vending variant at the end. Complete start to 
finish, no cross referencing, which is the point you made.
   
   The pages went from around 550 lines each to around 190, entirely by 
removing content that fileset-catalog.md now owns. Slugs are unchanged, so no 
inbound links break.
   
   The fileset catalog index page is not coming back. Its only content was a 
list of links to the five pages, and the Storage Backends table in 
fileset-catalog.md now links to each of them directly.
   
   One thing worth flagging that I found while doing this: the Python client 
spells the storage properties with underscores where the catalog and the Java 
client use hyphens, s3_endpoint against s3-endpoint. The old pages had both 
spellings in different sections without ever saying why. Each page now states 
it explicitly.
   
   The sidebar entries need restoring on the gravitino-site side, which is a 
separate repo and a separate PR. Note that fileset-catalog-with-cos was never 
in the sidebar to begin with, so that one is a fix rather than a restore.
   
   Ready for another look when you have time.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to