chore: Remove ikv online store - #6033
Conversation
![devin-ai-integration[bot]](https://avatars.githubusercontent.com/in/811515?s=60&v=4){kind=small}
| | support for entityless feature views | yes | yes | yes | yes | yes | yes | yes | yes | yes | no | | ||
| | support for concurrent writing to the same key | no | yes | no | no | no | no | no | no | yes | no | | ||
| | support for ttl (time to live) at retrieval | no | yes | no | no | no | no | no | no | no | no | |
There was a problem hiding this comment.
🔴 Incomplete IKV column removal leaves markdown table rows misaligned
The IKV column was removed from the table header (line 37) and from the first 9 data rows (lines 39-47), but the last 7 data rows (lines 48-54) still have 11 columns instead of 10. This causes the Milvus column data to be misaligned.
Root Cause and Impact
The header defines 10 columns: (empty) | Sqlite | Redis | DynamoDB | Snowflake | Datastore | Postgres | Hbase | Cassandra | Milvus. However, lines 48-54 still retain the old IKV column value, giving them 11 columns.
For example, on line 48:
| support for entityless feature views | yes | yes | yes | yes | yes | yes | yes | yes | yes | no |
The 9th data value (yes) was the old IKV value, which now incorrectly maps to the Milvus header. The actual Milvus value (no) becomes an orphaned extra column. This affects the rendered documentation for all 7 remaining rows:
support for entityless feature views: Shows Milvus=yes (should be no)support for concurrent writing to the same key: Shows Milvus=yes (should be no)support for ttl: Shows Milvus=no (happens to be correct by coincidence)support for deleting expired data: Shows Milvus=no (happens to be correct)collocated by feature view: Shows Milvus=no (happens to be correct)collocated by feature service: Shows Milvus=no (happens to be correct)collocated by entity key: Shows Milvus=yes (should be no)
Impact: Incorrect online store capability documentation for Milvus, leading users to believe Milvus supports entityless feature views and concurrent writing when it does not.
(Refers to lines 48-54)
Was this helpful? React with 👍 or 👎 to provide feedback.
Signed-off-by: tokoko <togurgenidze@gmail.com>
Signed-off-by: tokoko <togurgenidze@gmail.com>
Signed-off-by: tokoko <togurgenidze@gmail.com>
There was a problem hiding this comment.
🟡 Bigtable content-ref in docs links to mysql.md instead of bigtable.md
The bigtable entry in the online stores README links to mysql.md instead of bigtable.md. This is a pre-existing bug that was not introduced by this PR, but is visible in the changed file at docs/reference/online-stores/README.md:34.
Root Cause
At line 33-35 of docs/reference/online-stores/README.md:
{% content-ref url="bigtable.md" %}
[bigtable.md](mysql.md)
{% endcontent-ref %}
The url attribute correctly points to bigtable.md, but the markdown link text points to mysql.md. Depending on how the docs renderer processes this, users clicking the Bigtable link may be taken to the MySQL documentation page instead.
Impact: Documentation navigation error — users looking for Bigtable docs may land on the MySQL page.
(Refers to line 34)
Was this helpful? React with 👍 or 👎 to provide feedback.
* chore: remove ikv online store Signed-off-by: tokoko <togurgenidze@gmail.com> * chore: fix online store overview Signed-off-by: tokoko <togurgenidze@gmail.com> * fix: pre-commit Signed-off-by: tokoko <togurgenidze@gmail.com> --------- Signed-off-by: tokoko <togurgenidze@gmail.com>
* chore: remove ikv online store Signed-off-by: tokoko <togurgenidze@gmail.com> * chore: fix online store overview Signed-off-by: tokoko <togurgenidze@gmail.com> * fix: pre-commit Signed-off-by: tokoko <togurgenidze@gmail.com> --------- Signed-off-by: tokoko <togurgenidze@gmail.com>
* chore: remove ikv online store Signed-off-by: tokoko <togurgenidze@gmail.com> * chore: fix online store overview Signed-off-by: tokoko <togurgenidze@gmail.com> * fix: pre-commit Signed-off-by: tokoko <togurgenidze@gmail.com> --------- Signed-off-by: tokoko <togurgenidze@gmail.com> Signed-off-by: Shizoqua <hr.lanreshittu@gmail.com>
1 participant
What this PR does / why we need it: