Age | Commit message (Collapse) | Author |
|
This adds a migration path forward and back for migration of data in
migration 0003 (add the LEV_DISTANCE field).
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
It's more useful to test this after data has been added as this will
cause more code paths to be tested as it will be rendering the test data
that has been added
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
Also added test for 200 on urls for the SimilarTo, Build and Details
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
This adds the feature to recognise if an error has been reported before.
It matches it against the distance with an offset of bytes by using the
first error in the database as the prime error. The accuracy is
adjustable by setting SIMILAR_FAILURE_DISTANCE. If using the JSON
response to a submission you will also get the number of the duplicates
and a link to view them.
[YOCTO #7798]
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
We don't want to word wrap here because we often have long version
names/git SHAs and urls.
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
Currently the branch, commit, recipe and recipe_version rely on being in
a particular format so that they can be extracted by a regex. If this
fails we should fall back and mark these fields just as "unknown"
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
This was accidentally renamed in refactoring the parser save routine and
failed silently
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
At some point this should be moved to the model definition
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
The new send-error-report client can ask the server for the max_log_size
so that it doesn't try to submit a log which is too large for it's
current settings.
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
Needed to do the before the test is run and check afterwards. Also
remove unneeded delete as the deletion will cascade to the foreign key
items.
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
Also update the error message when we can't parse the json payload
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
To enable us to drop support for the URL encoded json payload and use
the more conventional non-encoded application/json payload we add a
switch on the client's http user agent, which will allow us to identify which
type of payload is being submitted and pave the way for future changes.
Also add tests to test this new path. Refactor tests so that they delete
the data after running to enable retrieval without having to know the PK
[YOCTO #7245]
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
This optional field is used to add link back information for errors
where the build information is accessible such as from a CI system.
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
This was stopping certain results from appearing
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
Fix the exception handling for page parameters that broken when changing
to using the value getters on the request object.
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
If you're hiding a column which as a filter by or order by set on it we
need to revert this order/filter so that you don't have to re-add the
column to get rid of the filter/order.
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
Simpler to use this than our own property fallback mechanism
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
Show the navigation on the statistics page.
Add a default state for mode in the default context so that we don't
accidentally compare between two unset variables
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
Previously we just happened to be returning a 500 on invalid data
because the argument for the status code was set incorrectly.
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
If the recipe_version isn't ellipsized then we don't need a tooltip to
show the full string.
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
Add most of the text fields to be included in the search results
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
As there are a few things we want to show/hide in the template based on
the results_mode pass this value in to the template rather than using
the path or query string to determine the page.
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
Add handling to invalid limit, order_by, filter and page parameters
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
We no longer need to pass the template name as a parameter to the
details url as it's in the view definition. Add an index url pattern so
that going to http://<host>/ works rather than having to know the
correct url.
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
If there are invalid characters in the BUILD payload return an error
message to that affect.
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
To run the tests ./manage.py test
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
When we catch these exceptions make sure we send back a useful response
and a http 500 status code
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
Also make sure the url pattern is properly terminated for the non-json
result
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
- Remove unused duplicated templates
- Remove unused imports
- Add comment to bug about POST request
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
This feature allows the definition in the settings of a special submitter to
add a tab in the UI which is similar to the latest errors page but all
queries on it are additionally filtered by the submitter being the name
defined in the settings. See commented example in settings.py
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
When no user sort has been applied show the sorted state of the page
as being sorting by decending submitted_on which is the default.
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
Remove getInfo as this was the python based data querying mecahnism that
is no longer needed. Also remove unused imports.
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
- Make sure the sort arrows are displayed when sort in progress
- Don't show filter icon on column that is currently being filtered
- Add min span width for the commit column to avoid button squashing
- Make the autobuilder page the main page
- Correct the page title copy
- Don't show all the available columns have a sane default
- Only allow some columns to be removed instead of all
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
If you wish to have the details of the submission returned in JSON
format then append /JSON/ to the submission url.
Also add a filter which allows you to view the recorded failures for a
specified build /Build/<id>/.
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
Newer versions of django handle application/json in a more correct way
by having the payload in the request body rather than a parameter to the
HTTP post request. So that we can continue to parse data correctly from
old send-error-report scripts we need to undo the url encoding of the
data.
see https://docs.djangoproject.com/en/dev/releases/1.5/
section #non-form-data-in-http-requests
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
Return a json formatted result to show the status of a error submission
including the urls for each of the failures logged.
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
- Rewrite the main views so that all the data manipulation is done in the
database rather than in python. This was causing large performance
issues.
- Add simplified client side logic in a separate file called main.js
- Simplify the url structure
- Disable the registration module
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
See 'manage.py migrate --help' for more infomation
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
The graphs were getting overly cluttered so reduce the number of series
to 6.
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
Previously using python-nvd3 we were constrained by the templates
distributed and the properties exposed by the python
bindings/abstraction for nv d3 which were incomplete. To avoid multiple
passes on the page (sever side and client side) we can just generate the
graph with the correct js/html in the first place.
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
Fixes RuntimeWarning: DateTimeField Build.DATE received a naive datetime
while time zone support is active.
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
|
|
[YOCTO #6542]
Signed-off-by: Roxana Ciobanu <roxana.ciobanu@intel.com>
|
|
Signed-off-by: Roxana Ciobanu <roxana.ciobanu@intel.com>
|