Forwarded from
- bug 708213 comment 29
- bug 710483 comment 1
- bug 710806

"""
ring_next introduces the concept of whether a certain line was already used, or not. I.e. when the terminal starts and the shell displays its prompt, ring_next points to the second line, implying that most of the screen is "unused". If a command outputs a screenful of newlines and moves the cursor back to the top-left corner, ring_next starts pointinig to below the bottom of the screen, implying that every row is "used" now, while in the mean time no user-visible change occurred.

I have a feeling that there are no sequence of events (stuff printed to the screen, or user-initiated events such resize or mouse highlight) that should be able to differentiate between these two cases. I mean, if any action behaves differently in these two cases, that's probably against the user's expectation.

I have a guts feeling that probably other terminal emulators don't even have this concept of "used" vs "not yet used" rows.

This suggests to me that forcing ring_next to point to the row below the bottom would decrease the number of independent variables and hence could make the code simpler. Or, this (along with ring_len) could be made a technical value private to ring.c, the rest of vte shouldn't have access to this.
"""

Also try Edit->Select all right after the terminal starts (only the first line is selected), or after some output was produced and then a "clear" executed (the whole screen is selected). The suggested change would make it always select the entire screen.