There were vector graphics screens with short to medium persistence, but they needed constant refreshing from a digital memory of the currently defined strokes.
Tektronix invented an indefinite-persistence tube; the phosphor wasn’t really indefinite persistence, but there was a special electrical screen behind the phosphor that got charged wherever the phosphor was written by the electron beam, and then was magically able to continually produce electrons which moved forward to the phosphor, thus maintaining the illumination wherever the beam had written. A quick pulse of the appropriate voltage erased the electrical screen and thus erased the image being produced on the phosphor.
Originally used in the Tek storage oscilloscopes ca. 1960, these tubes were later used in a series of vector-graphic computer terminals and small standalone computers.
I played a very early “lunar lander” on a Tek 4051, which had a Basic interpreter in it. The lunar landscape would be created at random and then written on the tube. The lander itself would be portrayed with a low-intensity non-storage beam strength and updated as fast as possible. Its position of course was computed in real time in accordance with physical laws and your inputs through the keyboard. Once you landed—crash or otherwise—it would be written permanently, until you asked for a new game or hit a manual erase.
We had a Tek "storage" scope that I think worked on the same principle. I remember asking their local sales engineer how their scope could "capture" nanosecond-range rise times, and he spilled the beans about how it wasn't digitally stored, but the voltage was stored on a grid that was read out onto the screen. Worked pretty well except as time passed it got a little ragged-looking, especially if you fiddled with the intensity dial. It also had a "persistence" dial that could help you keep the image on the screen until you got your scope camera set up to take a picture of the display. Them were the oooooold days...