release: 0.43.1 - First Impressions (live-usage batch #29-#40)#288
Merged
Conversation
added 4 commits
July 22, 2026 09:34
Three defects in StepRun: - the API's input field is 'input'; the wizard sent 'inputs', which was silently dropped - the run started with no input data and failed required-field validation, so the user's first workflow 'did not start' - the template YAML ran inline and was never saved, so the first workflow also never appeared on the Workflows page; it is now saved via POST /workflows before running (best-effort - an existing name is fine) - /workflows/run returns 202 queued; the wizard pretended the queued run had already completed. It now polls the run to a terminal state (3s interval, 180s budget) and shows the real output or error.
Batch of live-usage findings #30-#40 from the DGX Spark box, developed and verified on-device while the local checkout was unavailable: - fresh installs start with an empty Workflows page: 17 bundled samples moved into the template hub (git mv to src/sandcastle/templates), the image ships support dirs under example-workflows/ - docker sandbox: line-buffer aiodocker's chunked follow stream - large step outputs fragmented and were silently dropped - steps record the actually-executed model (resolved), not the declared one - sandbox health uses the configured backend - no more permanent 'Runtime is down' on docker boxes; insight links to /system-health - workflow health badges judge the last run first, lifetime rate second - Czech diacritics in PDF exports (DejaVu in image, conditional latin-1 squash, renderer x-position/width guards) - header Spark badge shows the detected GPU name (NVIDIA GB10) via /api/runtime spark_gpu - version 0.43.1 + CHANGELOG
Three of the moved samples duplicated existing hub templates by declared name (text-summarizer -> summarize.yaml, competitive-intelligence-radar, AI_Brand_Sentinel -> ai_brand_sentinel.yaml). The hub validation caught the last one via a checksum mismatch: extracting the moved copy on a case-insensitive filesystem silently overwrote the existing lowercase file. Duplicates removed, the original restored byte-exact.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
0.43.1 - "First Impressions"
Live-usage findings from the DGX Spark box, each developed AND verified on-device (the local checkout was TCC-locked, so this branch was assembled from the verified gx10 working tree). Includes #287 (wizard first-run fix).
Changed
git mvtosrc/sandcastle/templates; hub 248 -> 265 on the test box); support dirs ship asexample-workflows/NVIDIA GB10) via new/api/runtimespark_gpuFixed
inputfield name, workflow saved to disk, queued run polled to terminal state/system-healthnim/ornithin DB, verified live)Verification
Every change verified live on GB10 (API probes, DB checks, bundle greps); CI carries the full python suite + dashboard suite + lint gate as the pre-merge verification for this batch.
Supersedes #287. Merging publishes 0.43.1 to PyPI.