GitLab

Some fixes there are necessary for client work depending on cubicweb 3.23.

--HG--
branch : 3.23

Closes #14789440.

--HG--
branch : 3.23

And thus consider `_forced_mode` first.
Related to #14789440.

--HG--
branch : 3.23

--HG--
branch : 3.22

--HG--
branch : 3.22

--HG--
branch : 3.22

Query the information schema to retrieve existing constraint (primary key and
unique) and drop them.

Thus don't use anymore DROP CONSTRAINT IF EXISTS (not implemented by e.g. old
versions of sqlserver). To do so a missing DISTINCT was necessary to avoid
duplicates.

--HG--
branch : 3.23

When one accesses the rql_varmaker, that's usually to use it. The pb was that when
the varmaker was already in page's data (which is stored as session data),
session storage such as redis won't see that the session data is dirty and has
to be stored back at the end of the request.

To fix this, systematically call set_page_data.

(grafted from 3432f0e2540d)

--HG--
branch : 3.22

When one accesses the rql_varmaker, that's usually to use it. The pb was that when
the varmaker was already in page's data (which is stored as session data),
session storage such as redis won't see that the session data is dirty and has
to be stored back at the end of the request.

To fix this, systematically call set_page_data.

--HG--
branch : 3.23

cubicweb/skeleton/debian/rules.tmpl is now cubicweb/skeleton/debian/rules

prevent lintian from complaining

--HG--
branch : 3.23

With the new package layout (everything under "cubicweb" package), the custom
install_lib rule which makes use of include_dirs defined in __pkginfo__.py did
not prepend the package name to source directories to be copied. Fixing this.

Also, in setup.py's export() function, the destination directories' path to be
created during source tree walk was wrong.

All this makes cubicweb/skeleton directory (which is not a package) properly
installed by setup.py.

The test in cubicweb/devtools/test/unittest_devctl.py wasn't properly
implemented because it used an installation of cubicweb in "develop" mode
which shadows such packaging issues. Also it used "python -m cubicweb" instead
of directly "cubicweb-ctl" and the former appears to fall back to using the
cubicweb package *from sources* instead of the installed one.

Now that this test runs against the installed version of cubicweb, fix
MANIFEST.in to include tox.ini files (cubicweb's and skeleton's) as this is
expected from the test.

Closes #14127941.

--HG--
branch : 3.23

Make debugging easier.

--HG--
branch : 3.23

--HG--
branch : 3.23

This might shadow so packaging issues.
And user can still pass --develop to tox.

--HG--
branch : 3.23

--HG--
branch : 3.23

--HG--
branch : 3.23

--HG--
branch : 3.23

Broken since source tree re-organization.

--HG--
branch : 3.23

Broken since 62a7100d774b.

Closes #13973741.

--HG--
branch : 3.23

--HG--
branch : 3.23

--HG--
branch : 3.23

Only superusers can set lc_messages variable. This prevents connection in many
cases.

Closes #14010345.

--HG--
branch : 3.23

--HG--
branch : 3.23

--HG--
branch : 3.23

--HG--
branch : 3.23

It seems to me that this is not used during Debian package build (tests are
not run), but should be needed at runtime with the cubicweb-dev package.

Also, at the moment, we have no pbuilder setting with this package available
and it blocks the release.

--HG--
branch : 3.23

Maybe this will make apycot work again...

--HG--
branch : 3.23

--HG--
branch : 3.23

--HG--
branch : 3.23

Yams 0.44.0 got released.

This constraints (on unique CWUser's login and EmailAddress's login) are
handled by the backend.

This follows "[hooks/integrity] kill CheckUniqueHook".

Extend test for the ValidationError along the way.

* split permissions test appart
* avoid usage of internal_cnx as this does not seem need