A calm start to the year – with meaningful improvements
The beginning of January is usually quiet.
We like that.
Instead of shipping big announcements, we focused on a few small updates that make TestSeed more reliable, more predictable, and easier to trust in CI/CD.
Nothing flashy. Just improvements that remove uncertainty from test data.
If you rely on deterministic data in your pipelines, these changes are worth a closer look.
Seed Versioning & Diffs: You Can Now See Why Things Changed
Seeds are no longer just “the current version”.
Every relevant structural change now automatically creates a new, immutable seed version, including:
- a stable hash
- a snapshot of the structure
- a clear, human-readable diff
You can immediately see what changed:
- fields added or removed
- required fields becoming nullable (or the other way around)
- enum values being extended
- constraint updates
When CI behaves differently, you no longer have to guess or dig through commits.
The seed tells the story.
Data Quality Modes, Fully Aligned and Deterministic
We also finished aligning all data quality modes under one consistent model:
- Valid stays predictable and boring
- Invalid uses explicit semantic markers for targeted error testing
- Mixed is valid-first, with exactly one controlled failure per entity
- Boundary is now strictly edge-valid, never invalid
There is no randomness hidden anywhere.
The same seed will always behave the same.
And that’s the point.
Reliability Safeguards You Rarely Notice (Until They’re Missing)
We tightened our regex handling to prevent performance and stability issues:
- unsafe patterns are blocked server-side
- slow expressions are detected early
- safe fallbacks prevent runtime surprises
These safeguards don’t show up in demos.
But they matter a lot once TestSeed runs inside real CI pipelines.
Why These Small Updates Matter
None of these changes are “big features”.
But together, they push TestSeed closer to its actual role:
A reliable source of truth for test data in CI/CD.
When teams start saying
“check the seed, not the test file”
we know we’re doing the right things.
Looking Ahead
With deterministic versioning and aligned data quality in place, the next step is making CI failures even easier to reproduce and understand.
We’ll share more soon.