If (part of) your work (like mine) consists of doing single shot measurements, then history doesn't matter. It is wasted memory because the previous measurement simply isn't relevant. So please give me the choice to have full memory in a single shot even though some or most of it doesn't show on screen right away.
Can you explain this a little better (not being argumentative, but I just don't understand)? What is it you are looking for that can ONLY occur after the moment of acquisition?
I need to take you a step back here.
First of all: When you measure something, you want to be able to make a relation between cause and effect so you can draw a conclusion from the measurement result. This may sound obvious but for many it requires training to measure / collect data in a way so it is possible to establish a singular cause & effect relationship.
Secondly: I'm not looking at something which can only occur after the moment of acquisition. I just want 1 acquisition with as much data as possible so I can see what happened before AND after an event. I might not always need the data but with shorter memory DSOs I have often wished to have more data in order to have more information about what happened before and after an error condition. That can show the problem so it can be fixed right away or can give valueable glues on what to set triggers for and thus save a lot of time (and money) instead of poking around in the dark. Keep in mind that my typical measurements are for DUTs which are highly sequential through hardware and/or software.
Back to the history buffer: When I'm debugging / testing embedded firmware / FPGA / hardware, each single shot measurement represents a new version with a fix or change. Which means that whatever is in the history buffer is no longer relevant information as I won't be able to match a specific acquisition with a specific change somewhere. There is no way to draw a meaningfull conclusion from the data in the history buffer. So I rather have no history buffer at all if that means getting more memory for the current acquisition.
A history buffer is only usefull if you can record all relevant information in/by the oscilloscope itself (input versus output) without making changes to the DUT. That way you can draw a meaningfull conclusion from the acquisitions in the history buffer. Keep in mind though that a history buffer and segmented recording are functionally equal it is just that an automatic history buffer is easier to use / setup.