Skip to the main content

cache invalidation · a personal blog by christoph lühr

Persistence

The second page of this capsule, and the reason it exists: one page cannot demonstrate a link between two of them, and a link between two of them is what the prefetch hint is for.

What the plate is a picture of

P3 is the phosphor that made amber monitors amber. Struck by the electron beam it emits at around 590 nanometres, and then — this is the part that mattered — it keeps emitting for a while afterwards, decaying to a tenth of peak brightness in something like ten to fifteen milliseconds.

That decay is what the plate on the other page plots: intensity against time, from the strike to about 16 ms. Most of the signal is spent in the first fifth. By the midpoint the trace has thinned to a scatter, and what survives to the right edge is a faint band along the beam axis rather than an image of anything.

Why anyone cared

A raster display redraws the whole screen fifty or sixty times a second. If the phosphor goes dark faster than the beam comes back round, the screen flickers; if it lingers too long, moving text smears. P3 happened to sit in the gap where neither is very noticeable, which is most of why it was chosen over the shorter-persistence P1 green.

The amber screen was never a design decision anyone agonised over. P3 phosphor persisted a little longer than green, flickered a little less, and cost about the same.

The rest was habit. Once enough terminals were amber, amber was what a terminal looked like.

Getting back

Back to the kitchen sink

That link is a plain relative gemtext target — ../gemtext.gmi — and the processor rewrote it to ../gemtext.html on the way out. Over Gemini this same file serves unchanged, and the link resolves to the .gmi it actually names.