If we trigger a deploy, then we want the deploy to actually run.
I am assuming that this was in place when there was a single GitHub
Actions workflow that would trigger this all the time, even when nothing
changed in `docs/**`. Since https://github.com/dagger/dagger/pull/1498,
we no longer have this problem.
After merging https://github.com/dagger/dagger/pull/1591, I was caught
by this behaviour which to me seemed surprising. This change should make
docs deploy via Netlify behave in a more predictable way.
Signed-off-by: Gerhard Lazu <gerhard@lazu.co.uk>
The goal is to capture the shape of the new docs. It is not meant to be
final, but it should be as close as possible. We only want the bare
minimum for new users that on-board with Dagger Europa. As soon as the
new europaSidebar replaces replaces the existing one, the previous docs
will still remain available - doc IDs are unique and permanent. We will
do this by simply changing the default `slug: /` to point to the Europa
Docs entrypoint, which is doc 1200.
Helpful Docusaurus link re multiple sidebars:
https://docusaurus.io/docs/sidebar/multiple-sidebars
The new pages are numbered from `1200` onwards. This is meant to reflect
the `0.2.0` Dagger version. This numbering felt more meaningful than
just continuing to increment existing numbers.
I didn't want to be "wasteful" with the digits and start at `2000`, but
that was my first instinct.
I am keen on getting this live on https://docs.dagger.io/1200/local-ci.
Anything that is not in production, is inventory. Inventory is bad.
The goal is to allow anyone that has a link to get a feel for the new
docs as soon as possible, so that we can all see how they improve in
real-time, and steer them continuously towards the desired state. We
should be aware of the timeline, and not muck about, but instead
evaluate constantly how close are we to "flipping the switch".
Remember, the best releases are those where switches are flipped (e.g.
`--europa)`. The feature will have been out there for weeks (maybe even
months), improved by talking to users and then one day realising that we
are done, and just enabling it by default. It's the same principle
behind these docs.
Signed-off-by: Gerhard Lazu <gerhard@lazu.co.uk>
- what_is didn't have an id, started with 1000 (it's easier to find this file now)
- operator-manual doc id was clashing with ci use-case
Signed-off-by: Gerhard Lazu <gerhard@lazu.co.uk>
* Remove mandatory 'cache' field
* New 'name' field to customize cache IDs instead
* Simplify and expand tests
Signed-off-by: Solomon Hykes <solomon@dagger.io>
* Script is written to filesystem, not inlined as argument
* Pass script either as string, or directory+filename
Signed-off-by: Solomon Hykes <solomon@dagger.io>
* `cmd` is renamed to `command` for readability
* `script` is removed. Feature moves up the stack (bash.#Run,
python.#Run)
Signed-off-by: Solomon Hykes <solomon@dagger.io>
This change "collapses" together log "groups" containing hidden fields
for the TTY logger.
e.g. `actions.foo._a` and `actions.foo_b` will show up as a combined `actions.foo`.
Caveats: Since we don't know in advance how many tasks are in a group, the state will change back and forth.
For each task in a group, the state will transition from computing to complete (e.g. blue to green), then back to computing and so on.
The transition is fast enough not to be visibile.
Signed-off-by: Andrea Luzzardi <aluzzardi@gmail.com>
