* pass GET slash command payloads through query string
Previously, both GET and POST requests received the payload via the
body, but this was incorrect for GET requests. Now, the payloads for GET
requests is sent via the query string.
* reorder tests for clarity
* switch command tests to use httptest servers
* restore original test command endpoints
* MM-9728: Online migration for advanced permissions phase 2
* Add unit tests for new store functions.
* Move migration specific code to own file.
* Add migration state function test.
* Style fixes.
* Add i18n strings.
* Fix mocks.
* Add TestMain to migrations package tests.
* Fix typo.
* Fix review comments.
* Fix up the "Check if migration is done" check to actually work.
* Implement basic scheme CRUD endpoints.
* Get All Schemes (Paged) Endpoint and store plumbing.
* Add get teams/channels for schemes.
* Fix unit tests.
* Review fixes.
* More review fixes.
* MM-10183: Adds channel scheme API endpoint.
MM-10182: Adds team scheme API endpoint.
MM-10182_3: Switch from scheme_id in path to body.
* MM-10182/MM-10183: Changes path from 'schemes' to 'scheme'.
* MM-10182: Fix merge error.
* don't report an error on plugin activation if already active
* improved plugin logging events
Log an error when a plugin's ServeHTTP fails, or when it unexpectedly
terminates.
Restart a plugin at most three times, allowing its failure to later
bubble up under the "failed to stay running" status.
* clarified plugin activation/deactivation
Avoid repeatedly activating when any configuration bit changes. Improved
logging.
* constrain plugin ids to ^[a-zA-Z0-9-_\.]+$ and enforce minimum length
Previously, the plugin id was used unsanitized to relocate the plugin
bundle, which allowed writing outside the `plugins/` directory by using
an `id` containing `../`.
Similarly, an empty string was accepted as an id and led to unexpected
error messages.
* remove plugins by manifest path, not id
If the id within the manifest ever diverges from the actual plugin
location, it becomes impossible to remove via the API. Instead, if the
plugin is found by id, remove the path containing the manifest.
* ignore plugins with nil manifests
If a plugin was detected, but had a manifest that couldn't be parsed, it
will be left nil but still be listed among the packages. Skip over these
in most cases to avoid segfaults.
* leverage mlog more effectively for plugins
* build issues
* Implementing structured logging
* Changes to en.json to allow refactor to run.
* Fixing global logger
* Structured logger initalization.
* Add caller.
* Do some log redirection.
* Auto refactor
* Cleaning up l4g reference and removing dependancy.
* Removing junk.
* Copyright headers.
* Fixing tests
* Revert "Changes to en.json to allow refactor to run."
This reverts commit fd8249e99bcad0231e6ea65cd77c32aae9a54026.
* Fixing some auto refactor strangeness and typo.
* Making keys more human readable.
* MM-10232: improve error handling from malformed slash command responses
Switch to json.Unmarshal, which doesn't obscure JSON parse failures like
json.Decode. The latter is primarily designed for streams of JSON, not
necessarily unmarshalling just a single object.
* rework HumanizedJsonError to expose Line and Character discretely
* MM-10259: pinpoint line and character where json config error occurs
* tweak HumanizeJsonError to accept err first
* Add Scheme model and stub store.
* Port ChannelStore to be Scheme aware.
* Make almost all the API/APP layer work with ChannelSchemes.
Only thing still hacky is UpdateChannelMemberRoles().
* Add basic SchemeStore implementation.
* Migrate UpdateChannelMemberRoles properly and fix tests.
* Update store tests and mocks so they work.
* Include creating default roles in Scheme create store function.
* Implement role deletion and start scheme deletion.
* Only use non-deleted roles for authorization.
* Add GetByScheme method to Team store.
* Add GetChannelsByScheme.
* Update store mocks.
* Implement scheme deletion in the store.
* Rename is valid function.
* Add offset and limit to queries to fetch teams and channels by scheme.
* Fix queries.
* Implement scheme awareness in Team store and add a migration.
* Tidy up ChannelStore mapping functions and add exhaustive unit tests.
* Add all missing i18n.
* Proper tests for TeamStore internal functions and fix them.
* Make additional TeamMember fields nullable.
* Make new ChannelMember fields nullable.
* Create new nullable columns without defaults.
* Make new fields in large tables nullalble.
* Fix empty list of TeamMembers.
* Deduplicate SQL queries.
* Fix spelling.
* Fix review comment.
* More review fixes.
* More review fixes.
* Add user.DeepCopy() function
* Add omit admins/non-admins to WS broadcast and use for updating users
* Updates per feedback and adding unit test for ShouldSendEvent
* Incorporate a Token into the invitations system
* Adding unit tests
* Fixing some api4 client tests
* Removing unnecesary hash validation
* Change the Hash concept on invitations with tokenId
* Not send invitation if it wasn't able to create the Token
* Fixing some naming problems
* Changing the hash query params received from the client side
* Removed unneded data param in the token usage
The underlying l4g library is not resilient to filter modifications in
the presence of concurrent goroutines. In particular, it's not safe to
call Close() on filters which might be actively held by a goroutine for
logging.
This change disables all modifications to existing filters once
initialized by the App layer. In practice, we might be able to get away
with some modifications to the existing filters (i.e. changing levels),
but the [golang memory model](https://golang.org/ref/mem) makes no
guarantees that it is safe to do so:
> Programs that modify data being simultaneously accessed by multiple goroutines must serialize such access.
We can solve this holistically by introducing the requisite locking
within our fork of the l4g library. For now, we just disable all
modifications.
