Now that the content in `guides` is all of the edx-platform docs, move
them all into the top-level docs directory to reduce confusion.
BREAKING CHANGE: Guides are now just docs. This will require updating
the publishing settings so that RTD looks for the conf in a different
location.
`module-js` and `module-descriptor-js` are old JavaScript group
indicators, left over from when we managed XModule assets via Django
Pipeline. We would like to get rid of them in order to make it easier to
build XModule JS without using Python.
There is one single usage of `module-js` in the entire platform (the
rest have been replaced with Webpack references, which is the
less-outdated way of managing XModule assets :). The lone `module-js`
reference was added in 2013 [1] so that circuit diagrams would display
in the course wiki. However, the ability to render circuits in the wiki
was removed in 2015 [2], so it is safe to remove the reference.
There is also one single usage of `module-descriptor-js`. It's in the
legacy bulk email editor, which hackily cribs from the old HtmlBlock
editor. Fortunately, we are able to simply replace the Django Pipeline
reference with the equivalent XModule JS Webpack bundle. (Note: The old
email editor is currently still supported, but is currently being
replaced by frontend-app-communications, so this hack will be gone
eventually).
Finally, this commit also sneaks in one styling fix: it adds the
HtmlBlockEditor CSS back to the aforementioned legacy bulk email page.
The missing CSS was causing a read-only 1-line codemirror editor to
appear below the HTML editor [3]. This bug was introduced during the
original XModule SCSS decoupling [4], which removed builtin block CSS
from the LMS-wide bundle, thus removing the HTML editor CSS from the
bulk email page. We imagine that nobody noticed because the bug only
exists in master (not Palm) and frontend-app-communications seems to be
globally enabled on edx.org. As a simple fix, we add the new CSS link to
the legacy bulk email page, and it renders fine again [5].
References:
1. 3fc59b3da5
2. https://github.com/openedx/edx-platform/pull/10324
3. Before fix: https://github.com/openedx/edx-platform/assets/3628148/25fc41b2-403d-4339-8c49-0b04664dfa02
4. https://github.com/openedx/edx-platform/pull/32018
5. After fix: https://github.com/openedx/edx-platform/assets/3628148/9a5d74f1-cc83-4ebe-8f0c-ee270f7721b8
Part of: https://github.com/openedx/edx-platform/issues/32481
We publish the two pages that were being published under here to the new
edx-platform docs under docs.openedx.org so update this doc site to
redirect to that site for all pages.
Once this version has been published, we should be safe to remove this
doc project from the platform and all future changes should go through
the single edx-platform docs site.
* get rid of six.text_type(s)
* get rid of six.b()
* get rid of six.string_types
* get rid of six.PY2/six.PY3
* get rid of six.iteritems() and six.viewvalues()
Reverts openedx/edx-platform#32717 since it is breaking the Docker build,
both in the edx-platform CI, and for Tutor Nightly.
edx@0.1.0 postinstall
scripts/copy-node-modules.sh
sh: 1: scripts/copy-node-modules.sh: not found
The problems seems to be that `npm install` is run before `scripts/`
is copied in, but the new post-install hook counts on
`scripts/copy-node-modules.sh` existing.
This reverts commit 4b64d8342d.
We ran into memory issues at RTD so skip the xmodule, cms and common
docs for now so we can get some of the stuff up. We can come back to it
later once we make things better.
During the review of ADR 17 [1], Régis pointed out [2] that the shell script
which replaces Paver's `process_npm_assets` could be automatically invoked as
an NPM post-install hook, ensuring that the step is seamlessly executed whenever
`npm install` is run. I had avoided using that suggestion, as I worried that it
would make it harder to move node_modules out of the edx-platform directory in
Tutor's openedx image.
Since then, two things have changed. Firstly, Tutor v16's new persistent mounts
interface [3] has lessened the importance of moving node_modules. Secondly, I
have realized that using a post-install hook would not preclude us from
modifying the underlying script (scripts/copy-node-modules.sh) to look in an
alternative location for node_modules, should that end up being something we
want to do.
This commit modifies the ADR based on those findings, stubs out Paver's
`process_npm_assets`, and adds the suggested post-install hook and replacement
Bash script.
References:
1. https://github.com/openedx/edx-platform/blob/master/docs/decisions/0017-reimplement-asset-processing.rst
2. https://github.com/openedx/edx-platform/pull/31790#discussion_r1122802492
3. https://github.com/overhangio/tutor/blob/master/CHANGELOG.md#v1600-2023-06-14
Part of: https://github.com/openedx/edx-platform/issues/31604
@feanil is overhauling the docs structure of edx-platform currently.
Recently, a change he made upped the docs build time from <15mins
to >40mins. Rather than revert the changes he is making, we're going
to take the docs build check out of CI for now. He plans to re-enable
the build check, but as a ReadTheDocs build, not a GitHub Action.
He's actively working on the docs, so he's not worried about any
regressions that may pop up while the check is disabled.
It's worth noting that the current docs check does not actually
publish its output anywhere.
* Revert "feat: remove `field-data` service from runtime initialization"
This reverts commit 6c435bb68c.
* Revert "feat: remove field data binding from the runtime"
This reverts commit 5f46ea52cd.
Since we want to replace the existing technical docs with this guides
project and we already build the technical docs into this project now.
Adding these redirects will allow it to be a drop-in replacement for the
existing published technical docs project.
We want to be able to move pages around and have sphinx automatically
redirect to the now location when that happens. This will allow us to
re-organize safely as we figure out the best way to layout the
documentation.
We are still seeing issues running devstack provisioning
of LMS because it's missing pkg-config. We believe this might
be the location we need to add the package to.
