openproject

Commit Graph

Author	SHA1	Message	Date
Oliver Günther	1e8c873544	Remove filesystem adapter This commit removes the entire FileSystem repository and adapter. It rids all specs (core and legacy) of remaining uses of this adpater and replaces it with a temporary subversion repository, where necessary. We also need to remove remaining filesystem instances from the database, otherwise users will receive errors when accessing said repositories on the Frontend. This does not destroy any data on filesystem, but database information on the repositories (remarks) are lost.	9 years ago
Oliver Günther	3377696563	Began frontend feature specs	9 years ago
Oliver Günther	2b39d8fbdf	Cleaning up, Rubocop	9 years ago
Oliver Günther	15d026081b	Convert repository fixtures for compatibility. Travis build system still runs Ubuntu 12.04., which includes subversion 1.6.17. Rather than introducing a whole new PPA to travis, this commit converts the new subversion test fixture to pre-1.6 compatibility. Also, OSX's bsdtar uses `._*` member files for special meaning, which will lead to filesystem specs reporting incorrect size on a Linux system. To circumvent this, filesystem fixture has been re-packed with `COPYFILE_DISABLE=1` (c.f., http://superuser.com/questions/61185)	9 years ago
Oliver Günther	abc597a6d2	First draft implementation of radio-based frontend.	9 years ago
Oliver Günther	05b27c1b98	Asynchronous creation, deletion of repositories from OP. This PR begins the push towards management of repository lifetime from within OpenProject. What this currently provides is the following: - A mixin `ManageableRepository`, included by `Repository::{Git, Subversion}` to provide an interface to repo management - Two services and delayed_job workers for creating (`CreateManagedRepositoryService`) and deleting (`DeleteRepositoryService`) - Customized rendering of subforms in the repository settings for arbitrary SCM vendors. TODOS: - [x] Provide sane ground to work on ( cf. PR #3149 ) - [x] Rework SCM settings form (`{scm}_tags` is currently HTML generated from within a Helper.) - [x] Ease integration of arbitrary SCM vendors - [x] Display meaningful error / warning when user looks at bare repository (currently outputs 500, as Git HEAD / svn revision is not available) - [x] I18n - [ ] Specs (in progress) - [x] Fix legacy specs - [ ] Decide what to do with `Filesystem` adapter. Open Questions: - Configuration: Currently replaces the existing options in `configuration.yml`. - Implementation specifics: - Sub-projects repositories reside in a hierarchy below their parents - How should the delayed_job react upon failure? (esp. creation of repositories) Relevant work package: https://community.openproject.org/work_packages/20416 Merge dependent on PR #3149	9 years ago
Oliver Günther	7ebac37c14	Repository Management - Refactoring and Preparation Related work package: [#20415](https://community.openproject.org/work_packages/20415). This is a combined refactoring and extraction effort of local SCM functionality for SVM, Git adapters in preparation for the explicit lifetime management of repositories in #20416 and implementation of (local) repository storage quota in #18393. This PR is still heavily WIP and for the time being, depends on the legacy specs for testing. To run the tests, execute the following steps: ~~~ rake test:scm:setup:all rake spec:legacy ~~~ I'm open for ideas on how we integrate scm adapters tests into current specs. --- - [x] Exchange `IO::Open` for `Open3` to avoid countless manual exitcode checks - [x] Extract local functionality in its own Module - [] Extend `LocalClient` module with links to create/delete services for #20415 - [x] Break down unbearable local parse routines (Hound will be over the place until fixed - but down from 260 to < 20) - [x] `entries` - [x] `revisions` - [] local repository factory - [] Specs - [] Fix / Modernize - [] More adapter-related specs	9 years ago
Jens Ulferts	a21335db36	safeguard spec	9 years ago
Jens Ulferts	c5de9a4fb7	move spec shared context up in the dir hierarchy	9 years ago
Jens Ulferts	751751f6b4	reuse safeguards for wp index loaded in specs	9 years ago
Jens Ulferts	d5f44a0c60	hoping to fix spec with the help of ng_wait	9 years ago
Jens Ulferts	52f6b69221	hoping to fix spec flicker It seems to fail more often around date changes	9 years ago
Jan Sandbrink	55a2cf9f51	replace references to Redmine::VERSION - we now live in OpenProject for quite some time - kept legacy specs because they are... legacy	9 years ago
Jan Sandbrink	8a0a60445b	update root representer - add it to the documentation - only provide links to resources that can be directly used (i.e. without templating) - provide useful general instance information	9 years ago
Jens Ulferts	1b9565afc5	reuse safeguards for wp index loaded in specs	9 years ago
Jan Sandbrink	992888df2a	add configuration endpoint to APIv3	9 years ago
Jan Sandbrink	68d521ca85	add link for adding attachments to work package	9 years ago
Jan Sandbrink	1f7485d5bf	fix HTTP 500 in WP form - occured when an invalid type was provided by the client	9 years ago
Jan Sandbrink	7e53b9ac04	remove spec expecting german translations german translations are not part of OpenProject anymore (they are now provided via Crowdin)	9 years ago
Jan Sandbrink	1b5bb5ba98	use AR errors as mock	9 years ago
Jan Sandbrink	88e052a53e	style [ci skip]	9 years ago
Jan Sandbrink	17940bf02c	change validation of WP children - do not add error on a symbol called :#123 (123 is the childrens ID) - this is against the semantics of ActiveRecord::Errors - creating symbols in a loop used to be a bad idea (might be better today) - it breaks error messages in the APIv3 (which is due to APIv3 working beyond the borders of AR errors, not the best of ideas...) - use :base instead, the place for unlocatable errors - include the fact that #123 refers to a child element in the error message Basically I feel that this makes the children validation cleaner and better. But somehow I fear, that there is some code out there that depends on the fact that error keys like :#123 existed...	9 years ago
Jan Sandbrink	bc30b28558	extract attribute name conversion into extra class - consider special renaming rules for well known attribute names -> currently covering wp attributes and createdAt, updatedAt	9 years ago
Jan Sandbrink	b374ebdd7f	store messages alongside error symbols - allows to skip translation part in ErrorBase - allows to have correct translations containing placeholder values	9 years ago
Jan Sandbrink	138d3dba29	fix some specs - error messages now contain punctuation - when :save returns false, there should be errors present - errors in form representers are now API errors	9 years ago
Jan Sandbrink	dcd2a3b03c	rework error instantiation - multiple errors are no longer a special case of specific errors - there is a separate error class to represent multiple API errors - convert ActiveRecord errors to API errors in a single place, ideally work with API errors from there on	9 years ago
Jan Sandbrink	c586a9cc8c	mock the model in model_contract_spec	9 years ago
Jan Sandbrink	13b9042293	treat all 422 errors the same for forms That means: Do render the form for all HTTP 422 errors, problems are highlighted within the form	9 years ago
Jan Sandbrink	8f110d8b93	adapt specs to new contract behaviour - also refine specs for done-ratio to check for non-fail when not changed - use shared examples for parent properties also in done ratio specs	9 years ago
Jan Sandbrink	ac9cffe741	remove tautology and add apropriate specs	9 years ago
Jan Sandbrink	e9059a97ee	style [ci skip]	9 years ago
Jan Sandbrink	ce0b906d91	dirty trick to allow using a Hash as mock also removes default value from initializer, which was never sane	9 years ago
Jan Sandbrink	0b7b4bf329	extract error merging into contract	9 years ago
Jan Sandbrink	218c75e765	fix update_by! user when not providing notes - previously the created Journal Entry would have been made by the specified user or the User.current, dependent whether :notes were provided or not - now the journal entry is always made by the specified user	9 years ago
Jens Ulferts	f2ae9be55c	increasing spec robustness, hopefully	9 years ago
peter	6a7ab0edb8	fix invalid response for /api/v2/authentication.json	9 years ago
Markus Kahl	bd0df07090	fixed specs	9 years ago
Alex Coles	8338b7a6e1	Update OP::Authentication to use uppercase headers Signed-off-by: Alex Coles <alex@alexbcoles.com>	9 years ago
Jan Sandbrink	2854584cb2	use more clear scopes for autocompletion lets see if that really makes stuff more understandable... I (personally) found the old definition of "all" to be too narrow. Now there is: scope == "all": I want to search over all WPs scope == "relatable": I want to search over all WPs that are relatable to WPs in the given project (via project_id) no scope: If I specified a project search over the WPs of that one, otherwise search everywhere (like "all")	9 years ago
Markus Kahl	c24c58a273	fixed specs; use default realm; test custom realm	9 years ago
Markus Kahl	bcd981df5c	consider X-Authentication-Scheme; opt scoped realm	9 years ago
Markus Kahl	d854df5175	suppress browser prompt on authenticate failure	9 years ago
Oliver Günther	e0af219a36	Fix invalid time sheet link This adds specs for top menu items	9 years ago
Jan Sandbrink	393fe4e70a	please the hound	9 years ago
Jan Sandbrink	3a30834ce0	tests for execute_as behaviour	9 years ago
Jan Sandbrink	f491d3d9ef	split specs for mail notification jobs per class	9 years ago
Jan Sandbrink	9eb7cfbe9b	actually allow the status to be set while moving apparently the status_id param was set to expect an array of values while being moved in this commit: `d7f399ab8d` Probably because the spec started to expect that when it was moved to be an RSpec test. I am not sure if this was ever a valid thing outside the tests...	9 years ago
Jens Ulferts	1fc2661375	another safeguard for the inplace edit spec	9 years ago
Evghenii Barhatov	a609680483	Allow raising exceptions from notification jobs	9 years ago
Jan Sandbrink	d3ceffddfc	actually allow the status to be set while moving apparently the status_id param was set to expect an array of values while being moved in this commit: `d7f399ab8d` Probably because the spec started to expect that when it was moved to be an RSpec test. I am not sure if this was ever a valid thing outside the tests...	9 years ago

1 2 3 4 5 ...

2662 Commits (1e8c8735443d6ad0c13d8cf3a9ab407cbdec2642)