This came up in the matrix Accessibility chat thanks to Peter. I'll summarise the chat: It is usual for a screen reader to announce the expanded state information after reading out the text content (which is the entire row). In principle a screen readers could make it a configuration option to announce this beforehand, but we're not aware of any. The old tree-grid in 102 had a separate column for the expanded/collapsed state, which [had accessible names](https://searchfox.org/comm-esr102/rev/5ed7e785b128e4bc9de4a2b4beff126eebbfa4f5/mailnews/base/src/nsMsgDBView.cpp#1732-1734). Basically the "Thread" column had three states: 1. "Expanded" when the row had expanded children. 2. "Collapsed" when the row had collapsed children. 3. Empty when the row had no children. This was maybe a slight misuse of the "Thread" column, since both the first two states are visually indistinct: both are the same same thread icon. I.e. visually the "Thread" column only distinguishes between "threaded" and "not threaded", but for screen readers it would also distinguish between the expanded and collapsed state when it was threaded. Note that the "twisty" arrow icon did not have an accessible name in 102 (as expected). As a result, screen reader uses (with the default view) in 102 would have had the row read out as "Expanded. My subject. person@email.org 01/01/2023, 10:00", followed by their screen reader reading out state information like the tree-depth, row number *and* the expanded state. This state information may be read out in a different tone or voice to distinguish it from the text content. However, since the default view had such a verbose ending with the email and date, I wouldn't be surprised if a lot of screen reader users ended the speech or moved on before the state information was read out, so the repeated expanded state wouldn't have been that prominent. In 115, the "Thread" column has no accessible name (and is explicitly hidden with "aria-hidden"). I think this is why this is being reported as a regression here: screen reader users from 102 would be used to this "Thread" column having this information being presented first.
Bug 1846257 Comment 3 Edit History
Note: The actual edited comment in the bug view page will always show the original commenter’s name and original timestamp.
This came up in the matrix Accessibility chat thanks to Peter. I'll summarise the chat: It is usual for a screen reader to announce the expanded state information after reading out the text content (which is the entire row). In principle a screen readers could make it a configuration option to announce this beforehand, but we're not aware of any. The old tree-grid in 102 had a separate column for the expanded/collapsed state, which [had accessible names](https://searchfox.org/comm-esr102/rev/5ed7e785b128e4bc9de4a2b4beff126eebbfa4f5/mailnews/base/src/nsMsgDBView.cpp#1732-1734). Basically the "Thread" cells had three states: 1. "Expanded" when the row had expanded children. 2. "Collapsed" when the row had collapsed children. 3. Empty when the row had no children. This was maybe a slight misuse of the "Thread" column, since both the first two states are visually indistinct: both are the same thread icon. I.e. visually the "Thread" column only distinguishes between "threaded" and "not threaded", but for screen readers it would also distinguish between the expanded and collapsed state when it was threaded. Note that the "twisty" arrow icon did not have an accessible name in 102 (as expected). As a result, screen reader uses (with the default view) in 102 would have had the row read out as "Expanded. My subject. person@email.org 01/01/2023, 10:00", followed by their screen reader reading out state information like the tree-depth, row number *and* the expanded state. This state information may be read out in a different tone or voice to distinguish it from the text content. However, since the default view had such a verbose ending with the email and date, I wouldn't be surprised if a lot of screen reader users ended the speech or moved on before the state information was read out, so the repeated expanded state wouldn't have been that prominent. In 115, the "Thread" cell has no accessible name (and is explicitly hidden with "aria-hidden"). I think this is why this is being reported as a regression here: screen reader users from 102 would be used to this "Thread" column having this information being presented first.