Description
Status | Subtype | Assigned | Task | ||
---|---|---|---|---|---|
Open | None | T215515 Termbox Editing Mode | |||
Resolved | None | T215951 Cancel edit mode to return to reading mode | |||
Resolved | Tarrow | T222167 Research end user effect of using one rollback approach or the other |
Event Timeline
TL;DR I can't tell
504857
507036
In both scenarios it is evident that the modification to the entity contained in the vuex store causes a continuous growth of JS heap used while the difference in implementation causes no observable or consistently measurable difference. Suggestion regarding methodology welcome.
Attempting to research with the following plan: Using the standard android SDK With a nexus 4 (google flagship phone from 2012) with and android API v22 (the oldest I could see in the list)
Connected to mediwiki-docker-dev by adding *.10.0.2.2.xip.io to the VIRTUAL_HOST ENV var of the web mediawiki-docker-dev container.
Then visiting default.10.0.2.2.xip.io to see the local copy of the site.
I then checked out each patch for investigation and manually copied to the wikibase view termbox submodule and ran the SSR server with docker-compose up.
Using a test entity of Q64 from wikidata imported into a fresh mediawiki-docker-dev database using WikibaseImport. The test entity was to be found on Q4 after import.
I was unable to determine any difference with Q64. Gifs of each codebase are attached. In each test the page was loaded completely and then refreshed entering edit mode and adding 4 x 2 characters to the English label before clicking cancel
I was unable to determine any difference with Q64. Gifs of each codebase are attached. In each test the page was loaded completely and then refreshed entering edit mode and adding 4 x 2 characters to the English label before clicking cancel