ALICE updates (to be made after testing)

[snblickhan]

Just want to start this thread off by saying this app is BEAUTIFUL and I'm so proud to be part of this team! Yes, there are fixes to be made, but I don't want to overlook the fact that the work that has been done so far is wonderful.

Now then! On to edits.

• [x] On Subjects page, change 'Consensus Score' heading to 'Number of Low-Consensus Lines' [REMINDER SB: update About text to reflect this change]

• [x] Possible to allow Esc to exit out of the text edit popup? e.g. I clicked on a line of text and want to quickly close it without having to click 'Close'. [approx. 1 day] (#184 )

• [x] on Help page, in section beginning with "To ensure correct indexing within ALICE": change Group_ID and Internal_ID to group_id and internal_id (remove caps)

• [x] Extract lines are not correctly rendered. They all appear to be the same coordinates as the reduction lines - Re: This may not be an issue since the first user to see a subject will put down the first lines. Subsequent users are likely to click on that original line and enter new text/agree with the original rather than drawing any new lines.

• [x] Add loading spinner to Edit Aggregation Settings modal (#172 )

• [x] Don't allow page reordering when subject is approved (#175)

• [ ] ~When viewing zoomed-in subjects and clicking on a line further down the page (i.e. outside the view port), the zoomed-in image does not automatically 'jump' to the selected line. Not sure this requires action/updating on our part (unless there's a simple fix?), but I can imagine users reporting it as a bug so thought I'd flag.~ (173)

• [x] Transcriptions can become mismatched when moving rows and clicking undo. Still investigating exactly when this happens (#176 )

• [x] Need an option to delete frames. Use case: if a 90 degree subject only has one transcription because someone drew a line backwards, Editors can delete the transcription on that page, but then an empty frame remains. [Becky is mocking up design for this]

• [ ] When moving from subject view to group view subject remains listed as LOCKED until page is refreshed. (Possible lag in unlocking -- call being made to unlock, but not instantaneous in TOVE?)

Feedback from BPL team:

• [x] Pan tool "a little clunky and gets stuck at times" [Will going to ask Shaun for feedback on this] (#183 )

• [ ] Need better signposting on the option to click back into Group list -- both BPL testers noted that it's not obvious how to return to the list of subjects when you're on the subject editor screen (pinging @beckyrother for input here).

• [x] It would be helpful to have the zoom/rotate option available while editing a single line. (#174 )

• [ ] ~It would be helpful to have the image display automatically scroll to the line being edited instead of using the pan tool to get to it.~ (173)

• [ ] Add 'delete subject' button to list page (Becky mocking up design & confirmation pop up)

• [ ] Need to display all pages in case of a page containing no transcriptions. Currently, only pages with transcriptions appear. If a subject has four pages and page four has no transcriptions, there is no way to access that page in ALICE.

Weird bug that might be tied to a specific edge-case subject

• [ ] Deleting multiple lines on a page caused some sort of crash? NOTE: This was a subject that had been known to take a long time aggregating because it includes printed text and columns in the middle of a page. Full error report below:
• I had just adjusted aggregation settings for subject 2 in TEST1 group (OPTICS --> DBSCAN). Then began reviewing lines, marking some as complete and deleting others. After deleting 2 lines and attempting to click into another one, the screen went grey and I received the following errors in the console:
• [mobx.array] Attempt to read an array index (10) that is out of bounds (5). Please check length first. Out of bound indices will not be tracked by MobX
• TypeError: Cannot read property 'length' of undefined
• Uncaught TypeError: Cannot read property 'length' of undefined
• DevTools failed to load SourceMap: Could not load content for chrome-extension://hdokiejnpimakedhajhdlcegeplioahd/sourcemaps/onloadwff.js.map: HTTP error: status code 404, net::ERR_UNKNOWN_URL_SCHEME
• DevTools failed to load SourceMap: Could not parse content for https://alice.zooniverse.org/static/js/main.b2591ace.chunk.js.map: Unexpected token < in JSON at position 0
• DevTools failed to load SourceMap: Could not parse content for https://alice.zooniverse.org/static/css/main.acb4dab1.chunk.css.map: Unexpected token < in JSON at position 0
• DevTools failed to load SourceMap: Could not parse content for https://alice.zooniverse.org/static/js/2.4d18aa89.chunk.js.map: Unexpected token < in JSON at position 0

[nciemniak]

Transcription remains locked when I am viewing the transcription and close the browser window. Update such that transcription unlocks upon window closing.

[mrniaboc]

Just had a play around with ALICE on ASM. It's awesome! But I have a few issues to raise:

1) The "Zooniverse ID" the subject ID? If so, why are they such small strings (only 2 or 3 letters)? Also, in the documentation it refers to this as "Zooniverse Subject ID" instead of "Zooniverse ID" which is the column header in ALICE. These should match.

2) The colours of the aggregated lines look awesome, but it's a bit weird that they don't seem to convey information. It feels like they are trying to tell you something about the aggregation, which is a bit confusing, because I don't think they are. I would suggested perhaps using the colours to convey the level of consensus.

3) When I'm zoomed-in on the image (which is really necessary) I'm having difficulties moving the subject around. Click and drag seems really flaky. Sometimes it drags, but often it starts to drag and then disengages. (I'm on Chrome Version 80.0.3987.132 (Official Build) (64-bit) / Ubuntu)

[wgranger]

@mrniaboc

1. The "Zooniverse ID" the subject ID? If so, why are they such small strings (only 2 or 3 letters)? Also, in the documentation it refers to this as "Zooniverse Subject ID" instead of "Zooniverse ID" which is the column header in ALICE. These should match.

These smaller subject IDs are expected if you were looking at the "TEST" group on ASM. This is because we inserted duplicate transcriptions/subject IDs into TOVE, but intentionally gave these duplicate subjects lower IDs (that we acknowledge don't match their true Zooniverse IDs) since they needed to be unique. This is a temporary fix.

[trouille]

In Chrome v83 on macbook pro mojave 10.14.3

I did a few test things (rearranging lines, adding lines, etc.) and then clicked undo ~4 times and ended up with a blank screen. Did I break something, or is this expected?

Minor: flag and mark as read icons are quite close and my cursor would do funny things (flickering) when going to click on one of them (maybe because of proximity to other)?

[wgranger]

@trouille There are a few known issues that still need to be sorted out with the undo button.

[wgranger]

• [ ] When viewing zoomed-in subjects and clicking on a line further down the page (i.e. outside the viewing field), the zoomed-in image does not automatically 'jump' to the selected line. Not sure this requires action/updating on our part (unless there's a simple fix?), but I can imagine users reporting it as a bug so thought I'd flag.

@snblickhan I believe the BPL team noted this as well. This was not part of the original functionality of ALICE. If I'm understanding this correctly, it is not the quickest fix and will probably take at least a couple days to do this correctly.

[beckyrother]

Deleting frames: We discussed on #177 that collapsing the filmstrip viewer might be unnecessary. If that's the case, we should use that space for the Delete Page button, which displays Xs next to the pages and allows the user to delete a page. A confirmation box appears before action is taken. https://invis.io/HAVICPG872T#/371813754_Filmstrip

[wgranger]

• [ ] When moving from subject view to group view subject remains listed as LOCKED until page is refreshed. (Possible lag in unlocking -- call being made to unlock, but not instantaneous in TOVE?)

It looks like the "unlock" call is made to TOVE on both page close and going back to the transcriptions index. There is a rare state where:

• Editing subject 1234 (locked by you)
• Click breadcrumb to go to transcriptions page (unlock transcription)
• Transcriptions page GETs a list of transcriptions
• Subject 1234 is in that list, but the unlocked call from earlier hasn't finished yet and 1234 is still shown as "locked" (which is why a refresh handles this)

That said, I think this is a rare case that is handled by a refresh, and not a priority fix.

Still investigating: perhaps there is a state where the window closes before the "unlock" call finishes? I haven't been able to recreate that yet.

Removing the "LOCKED" text if you're the locker is another option, but I'm hesitant to do that as it could hide potential bugs in the future.