Skip to content

Latest commit

 

History

History
111 lines (99 loc) · 5.24 KB

Hub.md

File metadata and controls

111 lines (99 loc) · 5.24 KB

CurvHub

If Curv is to have an active community, then we'll need an easy way to publish curv modules for others to use, and for others to find a published module that meets some criteria.

Google isn't good enough, unfortunately. Some actual web infrastructure is needed. A searchable online index of published Curv modules.

It should have characteristics of:

  • shadertoy.com
  • thingiverse.com I.e., it's a place to publish and advertise cool shapes and parameterized shapes that you have created. When you do a search, you see a picture of each shape in the search results.

It should also have the characteristics of the online package repositories associated with all of the top programming languages. It's a place to find code libraries.

Curv does have a curated standard library that's built in to the distribution. But this doesn't need to be huge. We probably want to avoid the Python "stdlib is where packages go to die" phenomenon. OpenSCAD also has some rather questionable built-in functions, which can never be fixed or removed due to backward compatibility. So most Curv library code will live in CurvHub, which is overall uncurated.

CurvHub requirements:

  • It's a searchable index of 3rd party Curv modules.
  • There's lots of metadata available:
    • A short name by which the module is referred to in conversation. (A valid Curv identifier?)
    • A globally unique hierarchical name which is used for referencing the module from within code. Not like Python pip where the package name and the module name can be different.
    • An icon or animated GIF.
    • a list of #tags or keywords,
    • a short 1-line description
    • a long description.
    • author
    • Licence. I want to restrict the licences to a dropdown list of open source licences which are mutually compatible. With a reasonable default.
  • Popularity data. Eg, you may get a lot of search results, but you can filter on popularity. So how do we collect those metrics?

    • of likes; registered users can click "Like" button.

    • of views, on CurvHub

    • of downloads, from CurvHub

    • of references from other modules indexed by CurvHub

    • of remixes (as reported by authors, as on Thingiverse)? It's better if

      remixes are done using references. But it's valid to copy and paste code from another module so you can rework it into something different.

    • of collections that contain this module (like on Thingiverse)

IDE-like features:

  • You can click on a search result and view the source code online.
  • You can customize the parameters and view the results online, like Thingiverse.
  • You can modify the code and view the results online, like shadertoy.
  • You can use the web site like an IDE, create a new module, edit its code, test it, publish it.
  • Maybe we don't want CurvHub to completely duplicate all of the functions of github. You ought to be able to just link to a module that is hosted on github, or anywhere else for that matter, and do all of the actual development over there.
  • You can and publish a collection of CurvHub modules. This shows people what you like/endorse. It also makes these modules quickly available in your code editor when you are building a new module online.

Effects on the Curv language (what is a module):

  • Suppose I want to publish a 'voxelize' module, with a cute icon. It maps numeric parameters and a shape onto a minecrafty version of the shape. Other people should be able to refer to this by hierarchical name or url, add it to their collection.

Forum features?

  • You can post comments about a shape, like on shadertoy or thingiverse. Logged in users only. A "flag" button to report inappropriate comments. Staff members to monitor the flags and exert editorial control? Oops.

Decentralization

How much of this can be decentralized/delegated? (Without forcing casual users to set up their own servers.) Maybe check out

  • JoinDiaspora.com ("decentralized facebook")
  • Synereo.com (vague and ambitious, decentralized Attention Economy network) Also, total vapourware right now.

Business Model

I like how ghost.org is a non-profit organization. Their software is open source, financed by the ghost.org "platform as a service", which is subscription based.

So what can CurvHub charge for, while still keeping all of the software open source?

  • 3D printing of your model (or, more likely, someone else's).
  • T-shirt printing, and all the other ways to convert your model to a physical object.
  • This sounds similar to Shapeways, but in that model, the artist keeps the model secret and charges for prints. I want to create an open source community of shared remixable designs, more like shadertoy or github.
  • Server usage fees? Doesn't make sense if you are using Curv as a hobbyist and giving away your designs. Github has private repos, though.
  • If I support private designs, how does that impact my goal to foster an open source community?
    • Is there actually an industrial use case for CurvHub? What do you get from privately hosting on CurvHub?
      • the github model: cloud hosting and team productivity
      • the Shapeways model: sell your closed source model through a CurvHub provided storefront. But that kills the opensource goal, nobody shares, everybody wants to cash in.