Roadmap
Built and working
Section titled “Built and working”| Area | State |
|---|---|
| 2D channel drawing, snapping, node editing | Done |
| Straight and bezier segments | Done |
| Dimension markers and lofting across chains | Done |
| Ports — syringe, luer, luer-lock, custom — with a real-part catalogue | Done |
| Mixing chambers, manual and auto-placed | Done |
Solid CAD preview through manifold-3d |
Done |
| Monolithic and two-piece splits | Done |
| Manufacturability, flow-feasibility and drainage checks | Done |
| Drainage vent planning | Done |
| Analytical flow estimates, heatmaps, per-segment tables | Done |
| Heat exchange between adjacent streams | Done |
| Seven starter templates | Done |
| STL, OpenSCAD, SVG, PNG and presentation-bundle export | Done |
| Save, open, draft recovery | Done |
| First-run onboarding and guided tour | Done |
Known limits, by design
Section titled “Known limits, by design”These are not bugs and they are not on a list to fix. They follow from decisions that are load-bearing elsewhere.
The channel graph is planar, at a single Z. Every channel is at one height. This is why the herringbone template is a planar chevron reinterpretation rather than the published staggered device, and why channels cannot cross over one another.
Flow is a one-dimensional network model. No CFD, no concentration field, no multiphase. See The limits of analytical estimates.
One chip shape. A rectangular block. Anything else means exporting OpenSCAD and continuing there.
Local, single-user, no backend. No accounts, no sharing service, no cloud storage. Persistence and data logic are kept behind clean interfaces so a multi-user layer could be added without a rewrite, but none is planned.
Things that would be worth doing
Section titled “Things that would be worth doing”Unordered, and none of them promised.
Measured-resin profiles. A way to record what a test coupon actually gave you, per resin, and switch between them. Currently you edit the printer profile by hand.
Channel-crossing at different heights. Would require extending the data model beyond a single Z, and would change the CSG pipeline substantially. It is the biggest single capability gap.
A real staggered herringbone. Depends on the above, plus sub-channel floor relief.
More templates. Particularly a droplet sorter and a cell-trapping array.
Import. Reading an SVG or a DXF as a channel path.
Better mixing guidance. The app knows the residence time and the channel width, which is enough to estimate a diffusion length and say whether a mixer is plausibly long enough. It does not currently say so.
Out of scope
Section titled “Out of scope”- CFD, or any spatially resolved simulation.
- Slicing. Use your printer’s slicer.
- General mechanical CAD.
- A hosted service, accounts, or collaboration features.
- Anything requiring a network connection at runtime.
Contributing to this
Section titled “Contributing to this”See Contributing.