Added

The contents can follow you down a long argument

A Contents section that scrolls away is no use part-way down an argument, so it now follows you: once you scroll past it, a compact bar appears at the top of the reading area with the premise count, and you can open it to pick a premise without scrolling back up. The argument itself never moves when the bar appears — it is drawn over the page rather than pushed into it. The pin button in the section's header turns this off and on: pinned, it follows you; unpinned, it stays where it always was. The choice lasts as long as you are on the argument.

A contents list longer than the section can show now scrolls inside the section instead of filling the screen, and its rows are a little taller and easier to hit.

Arguments now open with a contents list

At the top of every argument, above the conclusion, a Contents section lists the argument's premises by name, numbered in the order you read them. Select one to go straight to that premise. If you had collapsed it, it opens for you. The section can be collapsed as well, and still shows how many premises the argument has. When an argument has counterarguments, the list follows the tab you are on.

Fixed

A failed import now says what went wrong more often

When an import fails, the page told you only what the whole run recorded. If nothing was recorded there but a step of the import had reported its own problem, you were left with a general "Import failed" line. That step's message is now shown instead, so the page names the cause in the cases where it previously had nothing to say.

An import's progress steps now say how each step ended

The row of steps on an import's page and on a building argument showed how each step ended only in the colour and shape of its mark, which a screen reader never read out. Each step is now announced with its own outcome — "Extract Claims, succeeded" — so the state is heard as well as seen.

Going to a claim inside a collapsed premise now opens it

Selecting a claim inside a premise you had collapsed, or pressing a passage in the source text that belongs to one, used to do nothing at all. The premise now opens and the page moves to the claim.

When something goes wrong on our side, you now get a message saying so

Some failures used to reach the app with no message at all, so the app showed a confusing error of its own or nothing happened. Others were reported as a problem with what you had entered when the fault was ours. Every such failure now comes back as a plain message that something went wrong on our end.

When the grammar checks refuse a change, the list of problems is shown again

If you had hidden the details of one refusal and dismissed it, the next refusal opened with its details already hidden, so all you saw was the count. Each new refusal now opens with every problem listed.

A change the grammar checks refuse is reported once

When the grammar checks refused a change, some actions showed a second message on top of the list of problems, such as "Couldn't update the premise" or a technical sentence under the claim editor, and some left an error in the browser behind. The list of problems is now the only report, and the form or menu you were using is left as it was so you can try again.

Some of those same actions did nothing visible when they failed for another reason, such as a lost connection: adding a premise, adding a claim from the claim search, and assigning or removing an axiom or a source. They now say the change didn't go through and ask you to try again.

When the change was made from a dialog, such as adding a claim or adding a source, the list of problems used to appear on the page behind it, where it was easy to miss. It now appears at the top of the dialog you are using, as it does in the app, and closing the dialog clears it. Adding a claim in a new formula, or answering a claim with a response, now lists the problems as well, where before it only said the change didn't go through.

Saving a source that couldn't be saved, for example with a required detail missing, used to do nothing visible. The form now says the source wasn't saved.

A failed quick import now stays on screen until you close it

When Quick Import by URL could not read a page, the message saying so disappeared after five seconds, so it was easy to miss. A successful import still reports itself and goes away on its own, now with a thin bar along the bottom of the message that empties as the time runs out.

A list of arguments that fails to load now says so, and offers Try again

When a page of arguments could not be loaded, a message appeared for a few seconds and the list quietly stopped working: nothing you did afterwards — scrolling, searching, changing the filters — loaded anything more. The list now says it could not be loaded, right where the rows are, with a Try again control, and everything works again once a page arrives.

A search you start while arguments are still loading now shows your search

Typing a new search while a page was still on its way could leave you looking at the older results, which arrived afterwards and replaced what you had asked for. The search you started last is the one you see.

"Nothing matches these filters" no longer appears while the list is still filling

On a long list, hiding some arguments with the filters could show that message for a moment before the rest of the arguments arrived. It now waits until there is nothing left to load.

Renaming, archiving or deleting one of your arguments now says when it fails

A refused delete looked exactly like a successful one — the argument stayed on the page with nothing to explain why. Every refusal now tells you what could not be done.

Discarding an argument waiting to be claimed now removes the one you chose

Discarding one of them left it on the page until you reloaded, and discarding a second removed the wrong one. Take ownership opened the argument whether or not it had been given to you. Discarding now removes the one you chose, a refusal says so and leaves it in place, and Take ownership opens the argument only when the claim was granted.

Starting an argument from a source that is switched off now names the source

The message named the source by an internal word — "Creating arguments from twitter is not supported yet" — instead of the name on the button you pressed.

Changed

The premise graph now matches the rest of Proposit, in light and dark

The graph you open with Show graph for this premise used colours of its own: a bright blue outline around the conclusion, a pure red one around a negated claim, blue and purple operator labels, and grey lines, background and zoom buttons that in dark mode sat as a darker box inside the window. They now come from the same palette as every other screen, and each one adjusts when you switch between light and dark. The lines between claims are also easier to see in dark mode. The conclusion is still outlined in blue, a negated claim in red, and AND and OR labels in blue; IMPLIES and IFF labels are now green.

Pressing Tab inside the graph also no longer stops on every box and every line between them, and the lines never showed which one you were on. It now moves only between the buttons and links you can use: each claim's sources, the zoom buttons, and Close.