TPAC 2016: automatic publishing

Why

The goal is still complete automation:
publish as part of your CI cycle

In the last 11 months…

(since Sapporo)

New types of spec supported

• ☑ Other types of WD's (eg, not on rec-track)
• ☑ WG notes
• ☐ FPWD

(Specberus supports most types, though)

Adapted to recent changes

• ☑ New /TR design
• ☑ /TR → HTTPS; Bikeshed & ReSpec updated too
• ☑ Supports 2015 process only (since 1 Aug 2016)
• ☑ Only HTML5 supported
• ☑ Specberus is the new pubrules checker; old one deprecated

Better UI in Specberus

• ☑ Including readable, linkable rule descriptions
• ☑ Simplified layout, more responsive
• ☑ Better code structure; more maintenable
• ☑ Better error messages (but WIP!)
• ☑ Links “see rule in context”, and pre-formatted GH issues

Specberus API & metadata

• ☑ Specberus REST API introduced; useful for CI (ReSpec?)
• ☑ Smarter metadata extraction
• ☑ …which enables automatic profile detection
• ☑ getdocrdf service, replaced by Specberus' API

TAR files accepted

• Result of discussions during/after Sapporo
• Has its own security mechanism
  (basic auth of user instead of shared token)

Not (yet)

• ☐ Support more types of specs
• ☐ Give truly user-friendly output
• ☐ Automatic checking of broken links/fragments
• ☐ Richer API

Priorities?

• ☐ Better UI
  • More descriptive error messages
  • Send better e-mail
  • More flexible (customisable) mail notifications
• ☐ Support more types of specs (FPWD, CR, submissions)
• ☐ Leverage our public API for more tasks
• ☐ Improve spec-generation (eg, unstability)

More ideas to prioritise

• ☐ Support publication directly from GitHub
• ☐ A web UI for Echidna (dashboard)
• ☐ Adapt to process 2016

⋮

• ☐ [What else?]

Public discussion

• spec-prod@
• public-pubrules-comments@
• #pub

⋮

• It's open source! Contribute!