I found Figma docs thrown over the wall from designers pretty hard to consume.

Perhaps it was just how that particular design team used it but

- they put a huge amount of information in a single document.

- lots of sticky note style additions that were not clearly associated with what they referred to.

- hard to search.

- hard to determine what a single unit of work could be. Again, maybe this was the team but they didn't have a concept of discrete tasking. It was just "here's the giant figma doc with everything from p0s to stuff we might not even want"

What are the upsides?

Interesting, we have a similar situation. A lot of information all at once, hard to navigate, search etc. Usually developers are so confused a figma doc is often used more like a prototype to show during a demo than a spec to share. Instead our designers make screenshots of it and put them in a well-organized Confluence document. I don't know what's the right way.

Figma will never be good at dev handoff because it isn't built on HTML/CSS rendering, so its code output is awful. No plugin or interpretation layer can fix this.

CSS is the language of design, yet for some reason every design tool completely ignores it.

Every top feature request on the Figma Forums is for HTML/CSS type features. We need a design tool built from the ground up on HTML/CSS rendering that has a Figma-like interface. Framer and Plasmic are sort of like this except made the mistake of wrapping everything in React code with tons of unnecessary abstractions that make the HTML/CSS output really messy.

I like the attitude relate.app (now Rainbow) has, but I don't see Figma being the final design tool, particularly when working with developers.

> CSS is the language of design, yet for some reason every design tool completely ignores it.

Maybe for web development, but people use Figma for mobile app designs too and those do not for the most part use CSS as the language for design.

But CSS would still get you better prototypes that imitate native designs than anything Figma can provide. All the top requested features for Figma are CSS-like features even coming from designers who don't know CSS. CSS has solved many of the most difficult layout and design problems, why reinvent the wheel?

I work on native app designs and other non-product design work in Figma and having the features CSS would provide out of the box would save me literally dozens of hours a month.

Yeah, auto-layout (flex) was weird at first but more and more designers get used to it, it becomes a preferable foundation for layout. I think they will love CSS eventually, it's just the teaching/learning process that can get them there. At least currently Figma hype gets them `display: flex`.

> every design tool completely ignores it

Figma shows CSS code in the Inspect view. It's limited, but not ignored.

It's not just limited, it is fairly misleading. I had to actively tell my devs to ignore the Inspect panel because it was giving them pixel values when we use rems, or they would hardcode colors that should have been design system tokens because Figma doesn't use color styles in things like effects. Lots of absolute values that should be relative or responsive or tied to tokens. There is a reason there are no good Figma to HTML/CSS plugins or tools out there.

Hi, I work on Plasmic. If not React, how would you encode composition, state, and behavior?

I wouldn't. A huge amount of interactivity can be done in pure CSS, and often design work is exploratory or experimental and doesn't need to output a fully functional component, it just needs the functionality CSS and HTML provide like interactive inputs and focus states.

Once you get the basic design abilities down, you can behind the scenes setup logic using some framework (sure you can use React, but I think tying a tool to a framework like that is a bad idea, React won't last forever and hopefully will die soon in place of newer better frameworks) that doesn't actually impact the basic HTML/CSS that is output.

Devs will still be needed to hook up designs to code logic, but at least give designers the ability to give devs clean HTML/CSS for them to wire up into JSX or Vue templates or whatever because currently many front-end devs are absolutely shit at HTML and CSS and React encourages lots of bad practices with HTML and CSS.

Some sort of open AST. Honestly "HTML" is pretty close except string typed attribute. `data-`|dataset is also pretty close.

> Some sort of open AST

well React is kind of that but it's not serializable. The "React Server Components" is trying.

---

We just need HTML but in form of AST so that it is serializable and programmable.

I don't follow how HTML comes anything close to expressing dynamism, composition, state, Turing completeness, etc.

Sure, you can encode whatever you want inside of data attributes, but that's just ignoring the problem. You still have to decide on a language to express all that within your data attributes.

React server components also does not is not trying to serialize this. Quite the opposite. We have an in-depth write up here: https://www.plasmic.app/blog/how-react-server-components-wor...

(Sise note: Plasmic's AST is in the open and has an API.)

I did a quick dig on Plasmic's AST and not found its spec. Is there a repo for its specification?

> 2. Server serializes root component element to JSON

That's what I'm talking about.

For scale: EV maker Lucid is worth 24 billion, and Apple recently had twice that in cash reserve. (Looked it up because of an unfounded rumor linking both.)

Yeah, I think the biggest improvement Figma made is that it made it so much easier to share and collaborate on artifacts from design with the engineering team. The previous process we used for taking Sketch designs and turning then into running apps was much more painful and cumbersome.

Except getting assets out of Figma is a huge pain. Trying to select the layers you need (if even possible), not being able to flatten layers into an exported image, not being able to compress the exported image...

Uhhh what? Sounds like you just throw everything in figma file without using artboards or components.

What do you mean you can’t flatten layers? Isn’t it what “exporting an image” is?

You can absolutely compress the exported image, well you can decide whether you want to export it at 1x 2x etc. It’s a bit different from changing dpi but honestly using figma for print is rough

Sometimes the artboards, components, or whatever the structure in use is doesn't match up with dev needs. Sometimes you need things split up, sometimes you need them flattened. It's just not flexible.

And 1x 2x etc isn't compression. I want a quality slider for JPGs, an 8-bit option for PNGs, I want to explicitly set the output resolution, I want to know how large the original source file is, how it's being cropped, etc.

"Sometimes the artboards, components, or whatever the structure in use is doesn't match up with dev needs" - this really sounds like you should tell your designers.

"how it's being cropped" - it's not being cropped.

"I want to explicitly set the output resolution" - why tho?

I ask because I found the lack of ability to set resolution as pretty jarring, however to my surprise I never needed it for digital assets

Figma can absolutely crop images, and sometimes you want to know how much of an image is actually available. And why wouldn't you want to set the resolution? Appropriately sized images are part of proper compression and performance tuning.

And you don't always have time to go back to the design team and ask them to rework tons of comps just because you need to get an asset out of a Figma.

I am not sure what you’re saying here about cropping. You can see how much of an image is available if you double click on same image. If your designer set it up in a billion of groups you may have to double click a whole bunch. But how is that different from cropping an image in illustrator? I’m not trying to be funny I genuinely do not understand what the complaint is here.

The resolution, I give you that, working with non-vector assets in figma is a bit annoying. But then again, I can export an image out of illustator at a 300dpi but if the original file was 72dpi, that will do absolutely fuck all.

As for not having time to go back to designers, I was not saying you do that every time, I meant, there is a way to have proper process when it comes to hand over to devs. Normally it’s a conversation you’d have once. If you work with people that have no regard for other people’s jobs, I don’t think it’s fair to be blaming the tool. I’ve seen absolute horror in design files on any platform.

Snap is losing money and doesn’t seem to have any path toward any decent profitability.

Spotify is a low margin business since no matter how much it makes in revenue, it will always have to give 70% to the labels - a company with high marginal costs that scales with revenue is the opposite of a successful tech company.

Cloud flare is also losing money.

Why would any of these companies besides Zoom be considered a successful company that should have a high valuation?