This project is mirrored from https://github.com/metabase/metabase.
Pull mirroring updated .
- Nov 16, 2021
-
-
Alexander Polyankin authored
-
Alexander Polyankin authored
-
Alexander Polyankin authored
-
- Nov 15, 2021
-
-
Cam Saul authored
* New logging utility functions for REPL usage * Test fix
* Add support for `:off` and `:fatal` levels. * Remove NOCOMMIT -
Dalton authored
-
Alexander Polyankin authored
-
john-metabase authored
-
dpsutton authored
* Get rid of the n-queries old logic: select last view logs, and then for each one, select its associated model (card, table, etc). new logic: get the last view logs, and do three batch queries, one each to card, dashboard, table, for the ids from the view logs * Mark datasets in api/activity/recent_views * Get rid of `(db/delete! Activity)` in tests * Handle datasets in activity list
-
Alexander Lesnenko authored
-
Ariya Hidayat authored
-
Alexander Polyankin authored
-
Howon Lee authored
Static viz waterfall needs BE bits. Here are the BE bits. Color and total summation coming...
-
Alexander Lesnenko authored
* add show total setting to the waterfall static viz * specs
-
Alexander Polyankin authored
-
Anton Kulyk authored
* Remove not used `noLink` prop * Namespace URL generators * Fix HeaderBreadcrumbs font-size in "head" mode * Add unit-tests for QuestionDataSource * Allow using react nodes in breadcrumbs * Use func declaration for QuestionDataSource * Add prop-types * Simplify query assignment expr * Don't swallow errors in QuestionDataSource tests * Add repro for #17961 (unit) * Allow passing custom divider for HeaderBreadcrumbs * Fix join tables breadcrumb links * Add extra saved question utils * Add tests for nested questions * Fix source question link * Don't show source question's schema * Fix breadcrumb colors * Fix tests * Fix breadcrumb colors * Use saved-questions lib utils for DataSelector
-
Pawit Pornkitprasan authored
-
Alexander Polyankin authored
-
- Nov 12, 2021
-
-
Dalton authored
* pass 'noEvent' api option to prevent unauthed event * unskip repro
-
Dalton authored
-
Jeff Bruemmer authored
* update sql parameter page * Flamber edits Co-authored-by:
flamber <1447303+flamber@users.noreply.github.com> * revisions * add mongo example Co-authored-by:
flamber <1447303+flamber@users.noreply.github.com>
-
Noah Moss authored
-
Alexander Polyankin authored
-
Noah Moss authored
-
Alexander Polyankin authored
-
- Nov 11, 2021
-
-
Jeff Evans authored
* Change query_cache.results column to correct blob type Add Liquibase changeSet to modifyDataType of query_cache.results to ${blob.type} (an idempotent operation on most DBs) Add schema migrations test to confirm, which simulates the broken app DB state, then runs the migration, then confirms the fixed type
-
Dalton authored
* update postcss deps * update the postcss config * replace color fn with color-mod fn * remove webpack-postcss-tools * fix postcss config * fix negated vars
-
dpsutton authored
* Speed up interger or string recognition jeff helpfully pointed this out. I did some repl benchmarks ```clojure (comment (doseq [x ["3" "bob"]] (dotimes [_ 3] (time (dotimes [_ 10000] (try (Integer/parseInt x) (catch NumberFormatException _ x)))))) (doseq [x ["3" "bob"]] (dotimes [_ 3] (time (dotimes [_ 10000] (if (re-matches #"\d+" x) (Integer/parseInt x) x)))))) tiles-test=> (doseq [x ["3" "bob"]] (dotimes [_ 3] (time (dotimes [_ 10000] (try (Integer/parseInt x) (catch NumberFormatException _ x)))))) "Elapsed time: 5.72825 msecs" "Elapsed time: 1.531042 msecs" "Elapsed time: 0.797041 msecs" "Elapsed time: 32.120875 msecs" "Elapsed time: 27.848375 msecs" "Elapsed time: 23.820542 msecs" nil tiles-test=> (doseq [x ["3" "bob"]] (dotimes [_ 3] (time (dotimes [_ 10000] (if (re-matches #"\d+" x) (Integer/parseInt x) x))))) "Elapsed time: 5.356417 msecs" "Elapsed time: 1.97225 msecs" "Elapsed time: 1.868708 msecs" "Elapsed time: 1.214666 msecs" "Elapsed time: 1.185625 msecs" "Elapsed time: 1.19025 msecs" nil tiles-test=> ``` * Only select lat/lon fields in api/tiles We required the index of lat and long columns, ran the whole query, and then just grabbed those columns in a `(for [row rows] [(nth row lat-idx)..])`. But of course we can just update the fields we want to select in the query. MBQL and native queries flow through this codepath depending on the source of the query of the card being mapped. For mbql queries, it is straightforward, add a filter on the lat long and replace the fields with just lat and long fields. For native, we nest the native query as a nested query and then proceed on the resulting mbql query. The only difference is requiring the type annotation for the field type. mbql field: [:field 32 nil] vs [:field "latitude" {:base-type :type/Float}] This addresses memory concerns. The query is not limited and we don't need to select an entire row for these purposes. Dropping lots of text fields, id fields, etc could save quite a bit of memory when resolving the entire result set in memory. * Add limit to number of coordinates per tile these queries were unbounded in getting coordinates for each tile. But in a visualization, can there really be more information provided? We were leaving them unbounded and getting back millions of rows (reported on #4844). So we were adding unnecessary detail at the price of OOM errors. Note tests have been changed to show that the limit of the original mbql query is clobbered and the limit of the original native query is only present in the nested query whereas the outer query selects its own limit. As always, should this number be exposed as a setting? Configurable in the UI? Leaving as a hardcoded and documented number for the moment. * docstring * Remove col indices from api/tiles since we just select the actual columns we want, we no longer need the index in the results of those cols. they are always the first and second columns * Trim final slash on url
-
dpsutton authored
* Nest qbnewb under /modal to allow for other modals move the api from api/user/:id/qbnewb -> api/user/:id/modal/qbnewb to allow for the upcoming new dismissable modal * Add additional datasetnewb route staying with the same naming scheme as qbnewb. the db field is is_datasetnewb, the route just takes datasetnewb. I'm not wild about the default is true, set to false, and the naming scheme. But will be easier to migrate and easier for FE to use if we continue the defaults/logic etc. * Remark on db change in migration file
-
Alexander Polyankin authored
-
- Nov 10, 2021
-
-
dpsutton authored
-
Jeff Bruemmer authored
-
Jeff Evans authored
Disable these tests for Redshift, since they are consistently failed with Out of Memory errors (#18935) * pivot-parameter-dataset-test * pivot-public-dashcard-test Re-enable if/when #18834 is resolved
-
Ariya Hidayat authored
-
Anton Kulyk authored
-
- Nov 09, 2021
-
-
Ariya Hidayat authored
Co-authored-by:
Maz Ameli <maz@metabase.com>
-
Alexander Lesnenko authored
* add waterfall static viz * review fixes
-
Anton Kulyk authored
* Slightly simplify QuestionDataSource * Extract and cleanup HeaderBreadcrumbs * Only keep database icon in breadcrumbs * Fix object ID not shown in detail view * Use `text-light` color for CollectionBadge * Support icon name as a prop of Badge * Add `inactiveColor` prop to Badge * Unify head and subhead breadcrumb components * Fix ad-hoc view items alignment * Fix invalid CSS * Use isStructured question method * Fix filters display in adhoc question header * Fix E2E selector * Fix saved question filters alignment
-
Ariya Hidayat authored
-
Jeff Evans authored
Secrets
PR 5 - Translate secret connection-properties into existing types understood by frontend (#18236) * Translate secret connection-properties into existing types understood by frontend For all :type :secret properties, before returning over the wire (i.e. from `available-drivers-info`), translate the connection-property into one or more new client side properties, using the existing types, to encapsulate the behavior of that individual secret "unit". Depending on the :secret-kind and hosting environment, this could mean there are multiple additional properties injected (one for the file upload versus local path option field, one for the value itself on upload, and one for the file path) Adding utility function to do the reverse (i.e. translating those client side properties back to the server secret property definition) Adding tests for both (including updating secret-test-driver to cover all of these) Add support for the treat-before-posting connection property config * Fix bug in `delete-orphaned-secrets!` -
Alexander Polyankin authored
-