Live data from Hacker News

Show HN: Thi.ng – open-source building blocks for computational design

news.ycombinator.com

21–30 of 43 posts

Re: Show HN: Thi.ng – open-source building blocks for computational design

#21
I've used Processing extensively, including toxilibs, and follow the creative coding community relatively closely. When I look at projects like openFrameworks or libcinder or Processing/p5.js, it's pretty clear to me both what those project are and how to get started with them.

I've kept an eye on thi.ng for years. I like the visuals on the former and current homepage and it seems like something that I would like to understand better. However, exactly what thi.ng is or how I am supposed to use it -- that always remains ever so slightly out of reach to me. The messaging is so focused on cobbling micro-libraries together that it feels like it is left to the reader to understand the possibility space of what you can achieve by doing so. For me, at least, that introduces enough friction to trying the tools that I feel tired and move on to something else.

If you can, I'd consider creating a few very straightforward Shiffman-esque tutorials (a la Nature of Code or something) that drop the discussion of why things are structured how they are and focus instead on helping the reader create something. I feel like if you help draw the line between the capabilities of the libraries and using them in production that the entire project will be more accessible.

Re: Show HN: Thi.ng – open-source building blocks for computational design

#22

Aah, thi.ng ! The creative coding framework I've been meaning to check out for years but never have! But the things (hah) made in it have always amazed me! So one bit of feedback on the site: it's fantastic as a portfolio site. However, while it's great for getting an impression for everything made with this framework and how large the framework is, what is missing is a big "START HERE" button. I get why that might f…

Thanks for the feedback - very much confirms some of my own issues, but it's also clear that I still have to find a better solution to emphasize some key points of differentiation: 1. thi.ng is definitely not a framework - it is a collection of 260 largely independent libraries & projects built around them 2. Because of #1, there cannot be a single "start here". I was hoping that the interactive tag cloud would make…

Well, I'd start with assuming the majority of visitors will expect a framework, and countering that with explicitly saying THIS IS NOT A FRAMEWORK! very early on on the website. Expectation management will go a long way.

Maybe a search box with "what topic would you like to play around with?", with some optional checkboxes to refine the search query (for example, maybe I want to work in the browser and limit my results to TypeScript and ClojureScript. That would also help a bit with getting the "toolbox" aspect across.

Another idea is to have example projects made with thing (the portfolio part) that also clearly demonstrate which of the libraries were used.

Re: Show HN: Thi.ng – open-source building blocks for computational design

#23
post #14

Earlier quoted context omitted.

If you read the comments in this thread it's clear that there are readers who haven't understood from the submission title or from the landing page what thi.ng is. When I read the word 'computational design' I don't immediately think of 'parametric design', 'algorithmic design', 'generative design'. Instead, I have to use google and read around. The landing page itself describes thi.ng as 'a truly manifold long-term…

That one is interesting indeed. I'm a bit curious about you taking issues with "manifold" and "polyglot" here: - manifold - something taking many forms - polyglot - https://en.wikipedia.org/wiki/Polyglot_(computing) I'm pretty sensitive to marketing BS/buzzwords myself, but unlike the usual "blazing fast" and similar hypephrases, the above two words actually are 100% correct use of the terms (IMHO) See my other reply…

Correct doesn't mean it's obvious to the average person, especially with uncommon words. I for one don't ever use manifold as an adjective. In fact, I only use it as part of technobabble nonsense to cheekily remind my mathematician friend that he's free to think that the topics he works on are trivial, but that the rest of the world won't agree with him.

[0] https://en.wikipedia.org/wiki/Manifold

Re: Show HN: Thi.ng – open-source building blocks for computational design

#24
post #5

I know you did a lot of work in Clojure(Script) originally, and IIRC you mostly switched to Typescript. Is that correct? How many of your projects are now designed to be consumed from CLJS vs JS/TS?

There're about 140 TS libraries vs. 70-80 Clojure/ClojureScript projects (around only half of which are libraries).

I partially stopped with the Clojure work because of lack of community/support, perf & workflow issues and changing interests (e.g. got back into C/Forth/STM32). Before that, I've tried for 4-5 years to encourage more people adopting CLJ/CLJS and to provide tools for graphics, data viz, digital fabrication etc. to make the language more appealing/useful for these folks. In hindsight it seems I gave up a year too early... Still, I find the TS monorepo workflow much more suitable for myself (and for the project), since large scale refactoring without types often has been a complete nightmare in Clojure (also using Literate Programming/org-mode made it even harder in these case). Plus, trying to write performant cross-environment code (for realtime graphics) in CLJC has been a quite painful/hairy experience. Still, as I said before, I do have very fond memories of the 6-7 years spent with the language & community. It's been partially commercial suicide (at the time), but an huge and important learning experience and I'm happy that (in the end) quite a few people found (and still do find) those tools useful...

Re: Show HN: Thi.ng – open-source building blocks for computational design

#25

Earlier quoted context omitted.

I don't think you miss anything. It is what it is, a collection of OS libs for generative art.

not OP, but the "generative art" part is what I missed

No, that’s what I missed as well.

Re: Show HN: Thi.ng – open-source building blocks for computational design

#26

Earlier quoted context omitted.

Thanks for the feedback - very much confirms some of my own issues, but it's also clear that I still have to find a better solution to emphasize some key points of differentiation: 1. thi.ng is definitely not a framework - it is a collection of 260 largely independent libraries & projects built around them 2. Because of #1, there cannot be a single "start here". I was hoping that the interactive tag cloud would make…

Well, I'd start with assuming the majority of visitors will expect a framework, and countering that with explicitly saying THIS IS NOT A FRAMEWORK! very early on on the website. Expectation management will go a long way. Maybe a search box with "what topic would you like to play around with?" , with some optional checkboxes to refine the search query (for example, maybe I want to work in the browser and limit my resu…

Ok, I think the issue is that people really just don't read anymore :)

First paragraph, second sentence:

"Not a framework, nor bound to any specific use case, environment or even language, it's a vast and mature set of complementing code libraries, which has organically grown to approx. 250 sub-projects"

As for the topic search: The search box says "search by topic" (fuzzy search) - the tags in the tag cloud are all clickable. Is that really not obvious (even though it's also mentioned in the text)? Click on "typescript" to only show typescript projects etc.

As for the example projects. This is also explained in the text. This work is ongoing. There're another 150 or so projects to be added (just my own). For community submissions there is & will be: http://awesome.thi.ng/ (currently just linking to the submission form)

Re: Show HN: Thi.ng – open-source building blocks for computational design

#27

Earlier quoted context omitted.

That one is interesting indeed. I'm a bit curious about you taking issues with "manifold" and "polyglot" here: - manifold - something taking many forms - polyglot - https://en.wikipedia.org/wiki/Polyglot_(computing) I'm pretty sensitive to marketing BS/buzzwords myself, but unlike the usual "blazing fast" and similar hypephrases, the above two words actually are 100% correct use of the terms (IMHO) See my other reply…

Correct doesn't mean it's obvious to the average person, especially with uncommon words. I for one don't ever use manifold as an adjective. In fact, I only use it as part of technobabble nonsense to cheekily remind my mathematician friend that he's free to think that the topics he works on are trivial, but that the rest of the world won't agree with him. [0] https://en.wikipedia.org/wiki/Manifold

Happy to entertain equally succinct counter proposals :)

Re: Show HN: Thi.ng – open-source building blocks for computational design

#28
post #21

I've used Processing extensively, including toxilibs, and follow the creative coding community relatively closely. When I look at projects like openFrameworks or libcinder or Processing/p5.js, it's pretty clear to me both what those project are and how to get started with them. I've kept an eye on thi.ng for years. I like the visuals on the former and current homepage and it seems like something that I would like to…

This is great feedback! I've got a lot plans for those areas, but the main roadblock is personal bandwidth. This has primarily been a spare time project for the past 4-5 years...

There're ~100 examples for thi.ng/umbrella alone, all commented, all linked from the readme's of each package used. Additionally, I've spent a lot of time this year adding more readmes, docstrings (again with examples), create diagrams etc. I completely get your comments/position, but I'm also really wondering if any of this work is just wasted effort if no one actually reads/looks at that stuff... and that then makes it even harder for me to prioritize.

FWIW I will also start streaming again in December, focusing on smaller projects/examples than previously. Maybe that format will work better for newcomers.

Obviously, there's no desire from my end to be overly cryptic about what you can do with these projects, but whereas P5, OF etc at their core all have a fairly narrow scope (DSLs for OpenGL API and some data wrangling sugar) and only gain their incredible flexibility through various addons and sheer community size, thi.ng simply has no core and that is the hardest thing to "sell" to people, even though IMHO it's also one of it's most unique aspects...

thi.ng is an ecosystem (as others have put it), not following the standard main pillar/silo + plugin architectures...

Re: Show HN: Thi.ng – open-source building blocks for computational design

#29

Earlier quoted context omitted.

Well, I'd start with assuming the majority of visitors will expect a framework, and countering that with explicitly saying THIS IS NOT A FRAMEWORK! very early on on the website. Expectation management will go a long way. Maybe a search box with "what topic would you like to play around with?" , with some optional checkboxes to refine the search query (for example, maybe I want to work in the browser and limit my resu…

Ok, I think the issue is that people really just don't read anymore :) First paragraph, second sentence: "Not a framework, nor bound to any specific use case, environment or even language, it's a vast and mature set of complementing code libraries, which has organically grown to approx. 250 sub-projects" As for the topic search: The search box says "search by topic" (fuzzy search) - the tags in the tag cloud are all…

> Ok, I think the issue is that people really just don't read anymore :)

The willingness to invest in reading many paragraphs of fine print (and really, as presented it is fine print) comes after a successful sales-pitch, not before it.

(Also, blaming bad UX on the user won't help in the long run)

> As for the topic search: The search box says "search by topic" (fuzzy search) - the tags in the tag cloud are all clickable. Is that really not obvious (even though it's also mentioned in the text)? Click on "typescript" to only show typescript projects etc.

It's not obvious in terms of telling anyone how this helps them get started. Sure I can search, but what am I searching through?

Re: Show HN: Thi.ng – open-source building blocks for computational design

#30
post #5

I know you did a lot of work in Clojure(Script) originally, and IIRC you mostly switched to Typescript. Is that correct? How many of your projects are now designed to be consumed from CLJS vs JS/TS?

There're about 140 TS libraries vs. 70-80 Clojure/ClojureScript projects (around only half of which are libraries). I partially stopped with the Clojure work because of lack of community/support, perf & workflow issues and changing interests (e.g. got back into C/Forth/STM32). Before that, I've tried for 4-5 years to encourage more people adopting CLJ/CLJS and to provide tools for graphics, data viz, digital fabricat…

I tried to get more people to pay attention to what you were doing, and still miss you in the Clojure community.
Post reply on HN