Documentation
Limitations
What runs where, the time budget, what the browser cannot promise, and what PPSCAD is not.
Everything runs in your browser
The code, the geometry kernel and the renderer run on your machine, in a Web Worker with no network. The server stores the code you save and nothing else; it never runs a model. A slow laptop makes a slow model, and a phone has less memory than a desktop.
A run has thirty seconds
A run that takes longer than thirty seconds is stopped and the Worker is started again. A model that hits the limit usually has a loop that makes thousands of solids or a sphere with far more segments than the print can show; lower the count or the segments.
Memory cannot be promised
A browser gives a Worker what it has and offers no way to cap it. A model that grows past what the browser allows ends the Worker; the IDE starts a new one and says so in the console, and your code is untouched. Keep the triangle count in mind: the viewer shows it under the model.
PPSCAD is not OpenSCAD
The two look alike and share most ideas, but an OpenSCAD file does not run as it is. Libraries written for OpenSCAD, minkowski, projection with cuts, surface, import of STL or DXF files and user fonts are not part of PPSCAD. The comparison page lists what to change, and the open specification says what PPSCAD is, statement by statement.
Text uses one built-in font
text() draws a plain built-in face: capitals, digits and the usual punctuation. There is no font choice and no kerning; for a label on a part it is enough, for a logo it is not.
Browsers
The viewer uses WebGPU where the browser offers it and WebGL2 otherwise, which every current browser does. The badge in the viewer says which one you have. Without either, the code still runs and the STL still downloads; only the picture is missing.
The draft lives in this browser
Until you save a design, your code is kept in this browser's local storage and nowhere else. Another browser or a cleared history does not have it. Save to keep it.