Tracking issue for a set of docs-repo infra changes developed on a
fork (netravnen/peeringdb-docs), proposed as a draft PR: #449.
This repo's README asks that issues generally go to
peeringdb/peeringdb#issues -- filing here instead since this is a
tracking issue for an open PR against this repo specifically, not
a bug report or support request. A lightweight FYI is also posted at
peeringdb/peeringdb#2027 for visibility on that side.
The underlying motivation: running a dev/preview fork of this site
surfaced three problems worth fixing generally, not just working
around locally -- preview deploys are indistinguishable from
production in a browser, blog and release-notes readers only had one
combined feed to subscribe to, and gh-pages accumulated deploy
history with no guarantee it matched the branch that built it.
Summary of scope (breakdown: #448 (comment)):
- A fork-preview banner (suppressed via
PROD=TRUE) so non-production
deploys of this site are never mistaken for docs.peeringdb.com.
- Independent Atom feeds for the blog and for release notes
(docs/blog/atom.xml, docs/release_notes/atom.xml), plus a shared
site-URL resolution policy used across the banner, both feeds, and
dynamic CNAME generation.
mkdocs gh-deploy --no-history, so the gh-pages branch is always
a single fresh commit rather than accumulating deploy history.
See the linked PR for the full commit-by-commit breakdown and open
questions for reviewers.
Tracking issue for a set of docs-repo infra changes developed on a
fork (netravnen/peeringdb-docs), proposed as a draft PR: #449.
This repo's README asks that issues generally go to
peeringdb/peeringdb#issues -- filing here instead since this is a
tracking issue for an open PR against this repo specifically, not
a bug report or support request. A lightweight FYI is also posted at
peeringdb/peeringdb#2027 for visibility on that side.
The underlying motivation: running a dev/preview fork of this site
surfaced three problems worth fixing generally, not just working
around locally -- preview deploys are indistinguishable from
production in a browser, blog and release-notes readers only had one
combined feed to subscribe to, and
gh-pagesaccumulated deployhistory with no guarantee it matched the branch that built it.
Summary of scope (breakdown: #448 (comment)):
PROD=TRUE) so non-productiondeploys of this site are never mistaken for docs.peeringdb.com.
(
docs/blog/atom.xml,docs/release_notes/atom.xml), plus a sharedsite-URL resolution policy used across the banner, both feeds, and
dynamic CNAME generation.
mkdocs gh-deploy --no-history, so thegh-pagesbranch is alwaysa single fresh commit rather than accumulating deploy history.
See the linked PR for the full commit-by-commit breakdown and open
questions for reviewers.