lakefs-client
is now deprecated, and we will shortly publish its last
release to PyPI.
Ever since we released lakeFS 1.0, we have supported these two SDKs to program for lakeFS with Python:
lakefs
, the lakeFS High-Level Python SDK. This is offers a structured easy way to use lakeFS from lakeFS.lakefs-sdk
, the lakeFS HTTP API. This is auto-generated from the lakeFS OpenAPI spec. It is guaranteed to support all lakeFS API actions. And it is covered by the lakeFS SDK interface guarantees.
These clients will continue.
Before lakeFS 1.0 we would publish only:
lakefs-client
, a legacy older version of the lakeFS HTTP API. This too was auto-generated. However it cannot provide interface guarantees. At no time could this client ever provide forward- or backward- compatibility at the source code level.
We have continued to publish lakefs-client
since the 1.0 release, in order
to provide continued support for existing users of this legacy Python SDK
client. But there are better alternatives, and its continued presence only
confuses our users.
Now it is time to deprecate this client, and we will stop publishing it at or after the version 1.40 release of lakeFS.
All existing uses will continue to work. Obviously we will not be pulling these published versions because of this change. We will stop publishing new versions of these clients.
Migration is covered by Migrating to 1.0. In most cases the changes are quite simple.
No.
Version 1.40 will be the last release of lakeFS to include a new version of
the legacy client. In order to use any new features of the lakeFS API,
you will need to upgrade your programs to use lakefs-sdk
.
Yes.
A program that uses any 1.x version of lakefs-client
still uses 1.x
versions of lakeFS APIs. Its usage of those APIs is still covered by the
lakeFS interface guarantees, so it will continue to work with future 1.x
releases.
That is also going away. However we see much less usage of it.
As always, our Slack htttps://lakefs.io/slack is the best place to interact
with the lakeFS community! Try asking on our #dev
channel