Standard Energy Efficiency Data (SEED) Platform™ is a web-based application that helps organizations easily manage data on the energy performance of large groups of buildings.
There was significant overhead to loading every django html request - e.g. just loading index.html on seedv2 takes 2.5 seconds before any data comes through or any other requests can be made, and this happens any time you load/refresh the page.
What's this PR do?
Completely removes the aforementioned overhead via COMPRESS_CACHEABLE_PRECOMPILERS - the problem was the npx sass command was running and generating a new css file on every single load, resulting in hundreds of thousands of files in the CACHE directory and multiple gigabytes. Now it only runs on change (500x faster).
Before
After
Overhead
2.5s
5ms
Enables Django compression using brotli, bundles all individual static files into logical groups, and only loads the files necessary for the specific context (so instead of making 251 requests on load now it only makes 23). Compression is only used in production, but you can enable it in a development environment. This change should also permanently fix stale browser cache issues (8x faster):
Landing only loads landing.css
SEED only loads vendors.css, seed.css, vendors.js, and seed.js
Before
After
Data Transferred
8.1 MB
1.9 MB
Load Time
1.71s
224ms
HUGELY improves performance when importing files with UBIDs into an org that already has records with UBIDs. For instance, if you import the 512-record covered-buildings-sample.csv to a new org, then upload it again with a single column changed and click Save Mappings (30x faster):
Before
After
Queries
798,549
33,142
Time
1h 57m 40s
3m 52s
Switches AngularJS tests to use Puppeteer instead of the deprecated Protractor+WebDriver setup (just run npm test, but keep in mind that it's expecting SEED to be running on localhost:80). This PR also switches to automatic test paths instead of hardcoded paths
Cleans up all landing files to remove unnecessary and (lots of) deprecated code
Fixes serving the FAQ page on Windows
Suppresses noisy Created ubid_model id logs
Sorts org members on the Accounts page in ascending order by 'last_name', then 'first_name', or otherwise email (previously the order was arbitrary)
Renames inventory_settings and inventory_detail_settings to inventory_column_list_profiles and inventory_detail_column_list_profiles respectively, since the previous names were very ambiguous
Makes the inventory list labels column pixel-perfect with borders consistent with the other fixed columns, and removes the broken sort button from the notes_count & meters_exist_indicator column headers
Formats large numbers with thousand-separator commas throughout the app, which is especially useful for orgs with large amounts of data
Makes clicking the SEED Platform text logo in the top left corner navigate instantly in all but one case
Updates frontend dependencies
How should this be manually tested?
Run pip install -r requirements/base.txt
To test using compression locally, modify your local_untracked.py file with the following setting:
Any background context you want to provide?
There was significant overhead to loading every django html request - e.g. just loading index.html on seedv2 takes 2.5 seconds before any data comes through or any other requests can be made, and this happens any time you load/refresh the page.
What's this PR do?
Completely removes the aforementioned overhead via
COMPRESS_CACHEABLE_PRECOMPILERS- the problem was thenpx sasscommand was running and generating a new css file on every single load, resulting in hundreds of thousands of files in the CACHE directory and multiple gigabytes. Now it only runs on change (500x faster).Enables Django compression using brotli, bundles all individual static files into logical groups, and only loads the files necessary for the specific context (so instead of making 251 requests on load now it only makes 23). Compression is only used in production, but you can enable it in a development environment. This change should also permanently fix stale browser cache issues (8x faster):
landing.cssvendors.css,seed.css,vendors.js, andseed.jsHUGELY improves performance when importing files with UBIDs into an org that already has records with UBIDs. For instance, if you import the 512-record
covered-buildings-sample.csvto a new org, then upload it again with a single column changed and clickSave Mappings(30x faster):Switches AngularJS tests to use Puppeteer instead of the deprecated Protractor+WebDriver setup (just run
npm test, but keep in mind that it's expecting SEED to be running on localhost:80). This PR also switches to automatic test paths instead of hardcoded pathsCleans up all landing files to remove unnecessary and (lots of) deprecated code
Fixes serving the FAQ page on Windows
Suppresses noisy
Created ubid_model idlogsSorts org members on the Accounts page in ascending order by 'last_name', then 'first_name', or otherwise email (previously the order was arbitrary)
Renames
inventory_settingsandinventory_detail_settingstoinventory_column_list_profilesandinventory_detail_column_list_profilesrespectively, since the previous names were very ambiguousMakes the inventory list labels column pixel-perfect with borders consistent with the other fixed columns, and removes the broken sort button from the notes_count & meters_exist_indicator column headers
Formats large numbers with thousand-separator commas throughout the app, which is especially useful for orgs with large amounts of data
Makes clicking the SEED Platform text logo in the top left corner navigate instantly in all but one case
Updates frontend dependencies
How should this be manually tested?
pip install -r requirements/base.txtlocal_untracked.pyfile with the following setting:What are the relevant tickets?
Tangentially related to #4549