Is this project still alive?

[Dublerq]

I haven't seen much activity in this project for months or even years. Travis checking is also broken. Is this still alive? I'm just wondering if it's worth to use it somewhere else than hello world test website.

[Inviz]

It's dead in the sense of progress, and it's dead in terms of usage too. I think the experiment proven itself not worthy - it's too heavy for regular sites. The company that developed it used it as a publication step, exporting final css, but that turned out to be kinda brittle. Most of runtime overhead is due to stateful nature of cassowary solver. Changing constraint set causes solved values to "stick", so gss does a bunch of work to handle and bring it closer to unachievable level of statelessness of CSS. It also juggles multiple solvers at once, to reduce size of constraint set in each solver to avoid exponential performance hit.

You still can use it (clone gss/document, npm link gss/engine optionally, npm install, and grunt build it), but you have to understand why you do it.

Good reasons:

• Using GSS.Engine as solver not bound to DOM (Sans GSS/document) - that's the best valid use-case for regular users, a wrapper around cassowary to deal with multiple solvers, unrolling constraints and doing computations off-thread.
• Learning how linear solvers work - completeness of constraint graphs, strengths and suggested values.
• Learning how to integrate multiple interpeters/solvers together - use static typing to analyze s-expression nodes, figuring out solving order, feeding results from solver to another, treating DOM as a solver.
• Integrating your own constraint solver to webpages - Bind DOM selectors and CSS properties to your own blackbox
• Using GSS as an interpreter for highly dynamic DSL - Using context-aware GSS interpreter to scaffold glue code to reactively execute your programming language, using Query as memory storage to track and clean side effects.
• Validating (box) model by expressing it through constraints - Figuring out minimal constraint and variable set that produces desirable results for languages that are not based on CSS

Bad reasons:

• Doing layout - just use flexbox
• Element queries - too heavy for the task, use some shim
• Hoping to use VFL as highly dynamic content-aware layout language - VFL is actually very rigid - it expects you to know what HTML you're getting to position everything, which is even worse than CSS.