Compatibility Lab Baseline Results — What We Published and How to Read Them

Walk through EditDocx’s public DOCX compatibility baseline fixture (SHA-256 included): which features passed, which were partial, and how to interpret the lab for real documents.

Marketing pages can claim “high fidelity.” A lab page has to show a fixture, a digest, and observations. EditDocx publishes that on the DOCX Compatibility Lab.

This article is the editorial companion: what the baseline fixture contains, what the 2026-07-16 result set says, and how you should (and should not) extrapolate to your own files.

Document workspace — use a real fixture, not vibes, when you judge fidelity

The fixture (reproduce the bytes)

Field Value
Name EditDocx compatibility baseline
Download /compatibility/editdocx-compatibility-baseline.docx
SHA-256 26bdee56fbbd33eeaf44f73e90502eb2626414a80a9d33eeab9e87225dc3e5d9
Contents Styled text, a table, an embedded image, a header, a footer, a comment, and tracked insert/delete revisions

If your downloaded file’s SHA-256 does not match, you are not looking at the published baseline — stop comparing notes.

Result matrix (published observations)

Status vocabulary: pass / partial / unsupported.

Feature Status Observation (lab)
Paragraph text and direct formatting pass Bold, italic, color, size, and alignment remain editable after import
Tables pass The 2×2 table opens with cell text, borders, and header shading
Inline images partial PNG remains in the DOCX package, but it did not render in the 2026-07-16 Chromium test run
Headers and footers pass Header and footer text render in their page regions
Comments and tracked revisions pass Comment and tracked insertion/deletion available to review workflow
Exact font metrics partial Font names retained; browser font availability can substitute glyphs and change line breaks

Summary from the published set: 4 pass, 2 partial, 0 unsupported on this fixture.

Lab date stamp: 2026-07-16 (source module). Engines and browsers change — treat the date as part of the claim.

How to read “partial” without panic

Inline images (partial) means: the bytes are still in the package (good for round-trip integrity), but the canvas did not show the PNG in that Chromium run. Downstream Word may still show the image after download. Always verify both on-screen and after download.

Exact font metrics (partial) means: the document still names the font, but your browser may substitute. Line breaks can shift. That is a browser-font problem as much as an editor problem — see the dedicated font substitution experiment.

What this lab does not claim

  • Pixel identity with Win32 Word for publisher templates
  • Support for charts, SmartArt, OLE, macros, or encrypted files (Features)
  • That every customer document will match these six rows

For failure-mode narrative beyond the fixture, read DOCX fidelity failures.

Reproduce in 10 minutes

Follow the step-by-step in Reproduce the DOCX round-trip test. Short version:

  1. Download the baseline DOCX and verify SHA-256.
  2. Open the editor and import the file.
  3. Check each row in the table above against what you see.
  4. Download a fresh DOCX and reopen in Word or LibreOffice.
  5. Note browser name + version next to any mismatch — that is how useful bug reports look (Contact).

FAQ

Why only one fixture?

The baseline is deliberately small and public. Expanding the corpus is ongoing; the lab page is the source of truth for published rows.

Can I trust a “pass” forever?

No. Re-run after major editor releases. The date on the lab matters.

Where do unsupported features appear?

They are listed on Features and in decision guides like when desktop Word still wins — not as silent failures on this fixture.