This:
1. Introduces a new override using the `pluggable_override` decorator.
It is now possible to specify a custom way of getting XBlock's icon
by defining `GET_UNIT_ICON_IMPL` in settings.
2. Introduces a way to add custom `XBLOCK_MIXINS` by defining
`XBLOCK_EXTRA_MIXINS` in settings. This allows, e.g. to add
new fields to XBlocks.
OeX_ES-44: Update edx-search elasticsearch.
* removes redundant doc_types from the lms/cms, replaces courseware
index with doc_types to two indices
* code polishing
* Use a new config variable for ES7 deployment. (#2226)
Co-authored-by: Golub-Sergey <1.golub.sergey.1@gmail.com>
Co-authored-by: Diana Huang <diana.k.huang@gmail.com>
Removed most of the deprecated shoppingcart app, leaving just enough to allow us to cleanly remove the related database tables later. Also removed the relevant Django settings that weren't in use elsewhere.
This uses the new names introduced in edx-django-utils
3.8.0 (edx/edx-django-utils#59), which we're already using, as
well as updating a few other locations where we incorrectly refer
to New Relic custom metrics instead of custom attributes.
Includes a couple of unrelated lint fixes in a file I modified.
The LEPL dependency was triggering a lot of deprecation warnings of the
form:
venv/lib/python3.5/site-packages/lepl/matchers/support.py:497:
DeprecationWarning: inspect.getargspec() is deprecated, use
inspect.signature() instead
argspec = getargspec(func)
It turns out that LEPL was only used by the rfc6266_parser package, which
itself was only used in one place to generate utf8-compliant
Content-Disposition headers.
This issue was noticed here:
https://github.com/SWW13/python-rfc6266-parser/issues/2
Unfortunately it is quite difficult to extract LEPL from the
rfc6266-parser package.
The rfc6266-parser package (https://pypi.org/project/rfc6266-parser/) is
itself a fork of the now-unmaintained rfc6266 package
(https://pypi.org/project/rfc6266/). Thus, it became high time to get
rid of this package. The FileResponse object can appropriately set the
Content-Disposition header, and thus replace the rfc6266 functionality,
since Django 2.0: https://code.djangoproject.com/ticket/16470
In our testing, the FileResponse object correctly set the
`filename*=utf-8''` value, following the RFC. The only difference is
that it does not provide "filename" fallback value, as expressed in the
RFC: https://tools.ietf.org/html/rfc6266#appendix-D
With rfc6266_parser:
>> import rfc6266_parser
>> rfc6266_parser.build_header("my_file_é.csv", filename_compat="video_urls.csv")
b"attachment; filename=video_urls.csv; filename*=utf-8''my_file_%C3%A9.csv"
With FileResponse we have:
>> from django.http import FileResponse
>> import io
>> response = FileResponse(io.StringIO(), as_attachment=True, filename="my_file_é.csv", content_type="text/csv")
>> response.get("Content-Disposition")
"attachment; filename*=utf-8''my_file_%C3%A9.csv"
We consider that this is a sufficiently minor difference, that will
impact very few browsers.
Cleaned up a few more things about the testing configuration now that we've deployed Django 2.2 to production:
* Go back to a single version of django-oauth-toolkit, tox doesn't need to vary it anymore
* Go back to PyPI release of django-method-override
* Remove no longer needed dependency on python-oauth2 (we already removed the package it was used by)
* Update prepare_xdist_nodes.sh with new Django requirements file options
This performs a dark launch compare of the existing implementation
(still in use) for fetching TPA provider configs and a new
implementation, recording metrics on exceptions and mismatches.
The new implementation should have two benefits, once we're switched
over:
- Fix 1+N queries on login page view where the site for each config
was fetched in a loop (ARCHBOM-1139)
- Don't allow configs with the same key on different sites to
interfere with each other (regression test added)
The new impl does not use TieredCache, but only the request cache,
which we may want to adjust later.
