Age | Commit message (Collapse) | Author | |
---|---|---|---|
2017-02-17 | [refactor] create resources only once | Kali Kaneko | |
it doesn't make sense to create the resources for every request, we can reuse the same resource and create it in the constructor. - Resolves: #8770 | |||
2017-02-15 | [style] pep8 | Victor Shyba | |
2017-02-15 | [tests] fix tests | Kali Kaneko | |
2017-02-15 | [docs] add a to-do about perf | Kali Kaneko | |
2017-02-15 | [feature] add robots.txt | Kali Kaneko | |
-Resolves: #6178 | |||
2017-02-15 | [feature] authenticate as anonymous if no token in header | Kali Kaneko | |
and serve / banner and robots to anon users. instead of returning 401 for all cases, I treat the unauthenticated case as a special case, and switch the service tree apart. this allows to serve a different resource tree to unauthenticated users. the new URLs are registered with the mapper. I don't really like that dependency, could be handled by twisted alone, but meh. - Resolves: #8764 | |||
2017-02-15 | [feature] add version to the banner | Kali Kaneko | |
2017-02-15 | [bug] add expected attribute to authentication wrapper in entrypoint | Kali Kaneko | |
the authentication wrapper is goin to look for the _credentialFactories attribute. it will raise an exception if not found - Resolves: #8766 | |||
2017-02-13 | [pkg] add script for deploying from git | Kali Kaneko | |
- Resolves: #8765 | |||
2017-02-13 | [bug] fix typo, again | Kali Kaneko | |
2017-02-13 | [bug] fix import typo on create-user-db | Kali Kaneko | |
2017-02-13 | [bug] get a new server instance on each request to the tokens db | Kali Kaneko | |
2017-02-12 | [refactor] update create-user-db script to use config refactor | Kali Kaneko | |
2017-02-10 | [bug] fix config handling after refactor | Kali Kaneko | |
2017-02-10 | [bug] revert loading from the wsgi entrypoint | Kali Kaneko | |
2017-02-10 | [pkg] use the entrypoint from the systemd script too | Kali Kaneko | |
2017-02-10 | [bug] effectively load the configuration for the app | Kali Kaneko | |
the code for passing the configuration to the couch initialization was never called. it seems the entrypoint module wasn't finally hooked as expected. I think this fixes the problem, but further review is needed here: either the entrypoint module is to be used, or it better is removed. in the first case, this workaround probably needs to be reverted. | |||
2017-02-10 | [bug] fix import for the session module | Kali Kaneko | |
2017-02-10 | [bug] fix typo in the resource path | Kali Kaneko | |
2017-02-09 | [pkg] add systemd service file to master | Kali Kaneko | |
this is to ease the packaging flow used in some environments like Pixelated, that use a debian branch against different branches. - Resolves: #8762 | |||
2017-02-09 | [refactor] parametrize blobs toggling in soledad server resource | drebs | |
2017-02-09 | [refacor] make proper use of twisted web dyamic resources in server | drebs | |
2017-02-09 | [refactor] allow passing threadpool pool for server sync resource | drebs | |
2017-02-09 | [feature] announce server blobs capabilities | drebs | |
- add a new ServerInfo resource for / - move entrypoint to its own module | |||
2017-02-09 | [refactor] rename server auth classes | drebs | |
2017-02-09 | [feature] add server config option for blobs | drebs | |
2017-02-09 | [feature] add server config option for blobs | drebs | |
2017-02-09 | [bug] Fix import for load_configuration on migration script | Thais Siqueira | |
2017-02-09 | [refactor] allow passing threadpool pool for server sync resource | drebs | |
Conflicts: server/src/leap/soledad/server/_resource.py testing/tests/server/test__resource.py | |||
2017-02-09 | [refactor] move wsgi sync setup to its own module | drebs | |
Conflicts: server/src/leap/soledad/server/_wsgi.py server/src/leap/soledad/server/entrypoint.py server/src/leap/soledad/server/resource.py testing/tests/server/test__resource.py | |||
2017-02-09 | [test] fix session and auth tests | drebs | |
2017-02-09 | [test] add tests for server auth session | drebs | |
2017-02-09 | [test] add tests for server auth | drebs | |
2017-02-09 | [refactor] remove twisted session persistence | drebs | |
The need for token caching in server is a matter of debate, as is the ideal way to do it. Twisted sessions store the session id in a cookie and use that session id to persist. It is not clear if that implementation is needed, works with future features (as multiple soledad servers) or represents a security problem in some way. Because of these, this commit removes it for now. The feature is left in git history so we can bring it back later if needed. | |||
2017-02-09 | [bug] fix name of module on import | drebs | |
2017-02-09 | [refactor] separate url mapper, avoid hanging tests | drebs | |
Because the wsgi resource has its own threadpool, tests might get confused when shutting down and the reactor may get clogged waiting for the threadpool to be stopped. By refactoring the URLMapper to its own module, server tests can avoid loading the resource module, where the wsgi threadpool resides, so the threapool will not be started. | |||
2017-02-09 | [feat] cache session data in server | drebs | |
2017-02-09 | [feat] use twisted web http auth and creds | drebs | |
2017-02-09 | [feat] reuse the url mapper instead of creating it for every request | drebs | |
2017-02-09 | [bug] disallow all requests to "user-{uuid}/" | drebs | |
2016-12-17 | [pkg] use a twisted resource as server entrypoint | drebs | |
2016-12-12 | [bug] enable batching again | Victor Shyba | |
Something happened during rebase. This configuration is supposed to be True by default now. | |||
2016-12-12 | [refactor] remove dead parameters, improve comments | Victor Shyba | |
received docs makes no sense for a single request download, plus all its comments and docstrings. Also updated docstrings for other methods. The method that tests if sqlcipher is encrypted can return a db handle that can be used right away. If we ignore it and reopen we can end up with a lost open cursor. | |||
2016-12-12 | [style] explicit unlimited request size | Victor Shyba | |
Request size on a stream can't be measured upfront and a limit doesn't make much sense. The real limit is user's Quota, to be implemented. | |||
2016-12-12 | [refactor] simplify server insert | Victor Shyba | |
Moved out magic numbers into a constant and simplified logic during doc upload. | |||
2016-12-12 | [bug] remove print debug statements | Kali Kaneko | |
2016-12-12 | [bug] fix bad merge in imports block | Kali Kaneko | |
2016-12-12 | [tests] adapt tests | Kali Kaneko | |
2016-12-12 | [refactor] adapt fetcher to decryptor | Kali Kaneko | |
2016-12-12 | [feature] batch based on payload size | Victor Shyba | |
batch is slower than usual insert for a single doc, so, if a document exceeds the buffer, commit the batch (if any) and put the huge load by traditional insert. refactor coming. |