Blog archive
All 50 posts, newest first. Browse the blog for cards and filters, or subscribe to the feed.
2026
Testaroo ships fixtures and gold kits so an importer can prove it works. Upload-ready means the importer already passed, not that a file merely exists.
Document Studio capability levels explain which edits are real, which are previews, and which jobs still need a dedicated tool. Ship-only language for authors and QA.
Clock skew across hosts, timestamps out of order, a stack trace spanning twelve lines, and NaN in a Prometheus gauge. The telemetry shapes that break ingest.
Units that differ by a thousand, buffers that live in another file, quantized vertices, and a node hierarchy 24 levels deep. What to check before a GLB hits the GPU.
Arabic has six plural forms, Polish has four, and most i18n bugs are a missing one. Eleven formats, the plural rules that break them, and what RTL actually requires.
Obfuscated fonts, an NCX that disagrees with the nav document, vertical Japanese, and a spine pointing at a file that is not there. The ebook cases worth owning.
Tokenisation that drops the term someone typed, ranking that buries the exact match, and zero results with no way out. The failures a search box hides.
Nested MIME, four charsets in one mailbox, authentication headers that pass and fail, and a bounce that looks like a reply. Test your parser against real message shapes.
Variable axes, COLR palettes, metrics that disagree, and a deliberately broken checksum. The font cases that break renderers, subsetters and export pipelines.
Antimeridian crossings, unnormalised longitudes, silent CRS assumptions, and a shapefile that is really five files. Test the cases that break importers.
Extractors fail on codecs they never linked, filenames they cannot decode, and nesting they never expected. Here is how to find out before your users do.
Assemble a compact editing regression pack from six-role audio and video workflow kits, expected analyses, controlled failures, and local browser-generated controls.
Turn a creator brief into a repeatable AI workflow test with provider-neutral plans, bound media inputs, expected outputs, and an explicit evaluation rubric.
Build synthetic media fixtures from versioned recipes, understand byte-stable versus content-stable exports, and verify every artifact without uploading source media.
Use one canonical field contract to test browser forms, paged layouts, structured values, field-data exchange, escaping, and round-trip imports across six form families.
Turn labels, autocomplete tokens, grouped controls, errors, and keyboard behaviour into explicit form-testing contracts using live demos and editable fixtures.
A practical guide to choosing between Testaroo's text editor, Form Studio, Document Studio, and Visual Template Studio without forcing unlike file models into one tool.
Use linked HTML, AcroForm, JSON, CSV, DOCX, XLSX, PDF, and Markdown artifacts to test field parity, local editing, export, and document workflows.
Build repeatable metadata-stripping and neural-network inference tests from paired controls, validated ONNX graphs, exact inputs, and published expected results.
Map Testaroo live form demos and HTML/AcroForm/FDF downloads to concrete autofill, validation, scraper, and PDF-filler assertions.
Concrete auth and PKI contracts you can assert against Testaroo SAMPLE JWT, JWKS, OAuth, certificate, htpasswd, and header fixtures.
A practical walkthrough of the Novus Visual Studio: change every line of copy, edit table rows, drop in your own logo or background, recolour a theme, and export the sizes you need, all locally.
SVG rendered through an img element runs in secure static mode and refuses every external resource. Here is why exports silently lose type and pictures, how to embed them, and how to test that it worked.
Turn Testaroo fixtures, filters, pairs, and spec sheets into a compact file-test matrix your team can rerun locally and in CI.
Use Testaroo Coverage, Glossary, Search, and Browse tools to choose a real test file by format, capability, and expected behaviour.
Use Testaroo live targets to test selectors, focus, frames, scrolling, delayed content, consent flows, uploads, themes, and print behaviour.
Use Document Studio to build a letter, memo, invoice, report, or résumé from editable blocks and export real Word, Excel, HTML, Markdown, or print-ready PDF, locally.
The Novus editor opens text and data files with syntax highlighting for 40+ languages, and receives binary files with a read-only hex and base64 view instead of a dead download link.
Use Form Studio to design a form, preview it live, and export standalone HTML, a JSON schema, or a real fillable-PDF AcroForm, with nothing uploaded.
Use Novus greyscale and sepia inputs against true-colour references to score colourization models with reproducible seeds.
Use paired clean, damaged, and mask fixtures to score restoration, inpainting, and damage-repair pipelines against ground truth.
Score watermark removers using clean references, synthetic SAMPLE overlays, and masks, without real brand logos.
Verify LUFS meters and clip detectors against amplitude ladders and clean↔hard-clipped WAV twins with exact specs.
Use Novus visual templates for marketplace listings, social posts, and product scenes, then customize them in Visual Template Studio without uploading to a server.
Exercise OpenAPI, JSON Schema, pagination, and webhook parsers against deterministic valid twins and intentionally invalid SAMPLE payloads.
How to evaluate speech recognition harnesses using Testaroo tone-digit WAV pairs and JSON transcripts, without copyrighted speech.
How to exercise autofill, scrapers, and PDF form fillers against fictional Testaroo tax, KYC, and shipping forms.
Score a point-cloud denoiser or reconstruction tool against a clean, dense ground-truth scan, with Chamfer distance and normal consistency.
Calibrate pixel and perceptual diffs against documented A/B twins: 1px shifts, chrome adds, and colour tints with known change kinds.
Test a mesh simplifier the right way: Hausdorff distance, normal deviation, and silhouette error against a dense ground-truth mesh.
Feed a broken mesh to a repair tool and score the result against a watertight target, with manifoldness checks and surface-distance metrics.
Score generated PBR materials channel by channel (base colour, normal, roughness, metallic, AO, and height) against a coherent ground-truth set.
Measure a depth model against a known depth map, with scale-invariant metrics, edge-aware checks, and reproducible fixtures.
Score a segmentation model with IoU, boundary F1, and instance matching against masks that come from the exact image you segment.
How to test ControlNet-style conditioning (Canny, depth, normal, scribble, segmentation, and pose) against a colour image the maps were derived from.
A repeatable way to measure denoise quality using noisy images paired with their exact clean reference.
Use one source exported across formats, plus a JPEG quality ladder, to test converters and compression.
Verify silence trimming against files with exact, documented leading and trailing silence timestamps.
Use a documented AcroForm and a text-vs-scanned PDF pair to test form parsers, fillers, and OCR accuracy.
Quoted commas, embedded newlines, ragged rows, odd delimiters, and encodings. The CSV cases every parser should survive.