Skip to main content

​
What’s new in Liberator 2.1

Liberator 2.1 brings two large workstreams together into a single release: enterprise SSO for the Liberator UI, and concurrent-query governance so a single user can no longer monopolize platform capacity. It also picks up a long tail of dataset-quality improvements — broader date-format coverage, a new file type, better handling of records containing special characters, and required timestamps on all queries.

​
Highlights

Enterprise SSO

The Liberator UI now supports standards-based single sign-on (SAML/OIDC), including federation with your existing identity provider and a customizable login page.

Concurrent query limits

Per-user concurrent-query caps prevent runaway clients from starving capacity.

PSV file support

Liberator now ingests pipe-separated value (.psv) files alongside CSV and Parquet, with the same date-format detection improvements applied to all file types.

​
Enterprise SSO

Liberator 2.1 promotes a standards-based single sign-on flow to the default authentication path for the Liberator UI:
  • Federation with your identity provider — your users sign into Liberator using their existing corporate SSO session. There’s nothing to configure on the user’s side, since CloudQuant wires the federation up for your tenant.
  • Customizable login page — the login page can be styled to match your organization’s branding rather than the default look.
  • Logout reliability — fixed a bug where clicking Log out would leave a stale session in some cases, causing the user to be silently signed back in on the next page navigation. The sign-out flow now consistently clears the session on both the gateway and the identity provider.

​
Concurrent query limits

A single user issuing dozens of long-running queries can no longer monopolize platform capacity. Liberator now enforces a per-user concurrent-query cap; requests above the cap are queued in the waiting room and processed FIFO. The cap is visible to administrators via the System Monitoring → Queue tab introduced in 2.2.

​
New file / data-format support

  • PSV (pipe-separated value) files — Liberator now treats .psv as a first-class file type alongside .csv and .parquet. The auto-detection logic mirrors CSV: header row, type inference, configurable null sentinel.
  • Additional date-format support — Parquet file uploads with date columns that previously failed during the date-processing step (mixed YYYY-MM-DD / MM/DD/YYYY / Unix-epoch columns in the same file) now ingest cleanly. The detection is per-column, not per-file.
  • Special-character handling in records — records containing characters that previously needed manual escaping (multi-line strings, embedded quotes, non-ASCII symbols) round-trip through the API unchanged.

​
Performance & reliability

  • Faster small-query response times — reused connections between Liberator’s query and cache tiers shave latency off small queries and reduce pressure on the cache tier under heavy concurrent load.
  • Better diagnostics on cache miss — when the platform misses cache on a symbol lookup it now records the requested symbol set, making “why is this symbol always being recomputed?” answerable from logs alone.

​
Bug fixes

  • Options dataset onboarding — a previously unsupported options intraday-history dataset is now onboardable through the standard ingestion path.
  • Pivot on uploaded files — pivot operations on Liberator UI–uploaded files that worked in earlier releases but stopped after a recent cut are working again.
  • Dataset descriptions show display names — the Liberator UI’s dataset description card now renders the dataset’s friendly display name instead of falling back to the internal slug.
  • S3-backed dataset date parsing — fixed a date-parse path that mis-handled certain S3-backed datasets with mixed date-column formats.

​
Subsequent fixes shipped in 2.1.1

Some issues that surfaced after 2.1 were fixed in the 2.1.1 patch, most notably a reliability issue under sustained SQL-source load.