> New users will not be able to sign up for Fig's products right now while we focus on optimizing them for existing customers and addressing some needs identified to integrate Fig with AWS.
This sounds a lot like the product is dead, and may emerge again at some point as an AWS hosted, Amazon branded product...
I'd never use a subscription + telemetry laden product like this in my core workflow, but sucks for the current users I guess.
To me the whole post read like "we are aquihired, but we cannot admit we are are aquihired". The closing of registration driving it home.
If AWS bought it for the product, what reason would there ever be to stop the current business model entirely rather than leaving it on its current trajectory for a few months, untill said "needs" are addressed?
Of all PR, I think I may hate acquisition PR the most. It is almost entirely just lies. And the same ones every time.
Just tell us if you just wanted the employees, just wanted some rights, wanted to integrate their product for your use, or just wanted to kill something off. I'm sick of reading the same new release for every acquisition knowing it's bullshit.
I read it as “hosted AWS version coming soon, we will email you when it’s time to migrate”.
If it was just acquihire there probably wouldn’t have a big blog post tying it to AWS. Just negative publicity for acquirer when they could shut down quietly.
I was at Cloud9 which was acquired by Amazon. We took most of the existing frontend and tech and rebuilt the backend on AWS, then relaunched it as AWS Cloud9. In the meantime we sunset the existing service.
I'd assume the same thing is going to happen here.
> If AWS bought it for the product, what reason would there ever be to stop the current business model entirely rather than leaving it on its current trajectory for a few months, untill said "needs" are addressed?
That’s an easy one: if it’s burning money like crazy on its current hosting, but only because of rents extracted by one of their competitors, not for any inherent OpEx-intensity; and that therefore Amazon expects it to be profitable when moved to live rent-free on their hosting.
I would never use a cloud-syncing cloud-connected terminal for simple security and privacy reasons alone, not to mention the fact that if it goes down I become basically disabled as a developer or admin.
Several companies have tried to SaaSify the terminal and failed, so I suspect I am not alone here.
"We've noticed that there's a component of the computing infrastructure that isn't sending everything you do to the cloud and charging monthly rent..."
Yes I'm genuinely shocked at the other comments here saying they use this happily; it should be a cause for concern. That it's now landing in Amazon's lap, doubly so.
If Fig goes down, you fall back to the regular shell that doesn't have autocomplete etc. If Fig gets hacked, that's scary, then again GitHub and others are similarly big targets.
Idk though, this doesn't seem useful enough to me as an individual that I'd bother looking into it. Maybe for a team or company.
As a side note, if your security relies on you typing things into your terminal, you're doing it wrong. AWS even makes you check a box when generating creds that basically says "I understand the way I'm doing this is insecure" when generating creds that will be used that way.
I would use this as the way for company's command line tools to be provisioned to me as a team member. This means my tools can be separated from work tools
It actually makes sense for Amazon integrating this with the whole browser based shell thing (which is a sensible secure default for a lot of people).
Having something with autocomplete that works seamlessly with their services seems like a better idea than a plain prompt if you're going to use their shell. Hopefully they do something sensible and customer-centric with the telemetry stuff, that seemed to be the big drawback for many.
These features are nice, but I've never liked the idea of having to use a whole new terminal application to get them.
I may be becoming a dinosaur, but it's not that I'm not willing to try new things. On the contrary, after many years of being rigid about having one true development environment, I've moved away from Emacs to VS Code, and work from more heterogeneous environments instead of being 100% Mac. So these platform-specific thick client apps no longer feel like the way to go.
Sticking to Vim and other very well-established CLIs is partially what enabled flexibility for me. As long as I have access to a *nix machine locally or over SSH, I have a capable dev setup that I'm well-versed in. Same thing for the past 8 jobs I've bounced around in, whether in-office or remote, with whatever security restrictions, with or without a desk, maybe with just a 13" laptop. It's the lowest common denominator. Meanwhile my teammates at my current job are constantly trying to protect their workstation setups so they can use their IDEs locally, and they've also been forced to switch IDEs twice in the past couple of years.
And this isn't coming from some FOSS ideology. I don't personally care about any of that, but the best tools are sometimes FOSS. I've got a Mac and can deal with Linux as needed.
To be honest I do think it's time we move on from emulating teletypes, but I also believe that whatever comes next has to be an open standard, with competing implementations.
There's a whole graveyard out there of "next-gen" terminal projects, and none of those still alive seems to be gaining dominance. I don't think writing down a spec would help this, but it would be great if whatever emerged victorious had one.
I think the part of this that feels wrong is that it steps into the realm of so many other perfectly good tools, and it seems like it’s there to enshrine crappy local development and deployment practices.
If you need reporting on the scripts your developers are running on their local machines, that probably means they have too much access to live environments. Guardrails and reporting should be in your CI/CD process and your source control.
Shells like fish or shell plug-ins already handle autocomplete nicely, so that feature isn’t that useful.
If you need env/dotfile management you probably need to spend time making a proper development environment rather than shimmying in a tool to make doing the wrong thing easier.
SSH key management is so trivially accomplished by numerous methods like configuration management or built-in solutions offered by cloud providers. Plus, most development teams almost certainly give out too much SSH access as it stands.
If you need to SSH regularly that means your application isn’t shipping logs and your local development tooling isn’t good enough.
Curious to learn more about your switch from Emacs to VSCode?
I'm a super confused soul, and currently I use all of them:
- Emacs (for clojure / lisps / big projects).
- NeoVim (for when I want to work in the shell, smaller scripts/playground, ssh).
- VScode (js / ts / frontend / figma / etc).
I've made both my vim and emacs keybindings / features almost identical, so there is limited context switching there.
I feel very slow in VSCode, I don't even have one of my most common task optimised in VSCode yet: grepping something and quickly moving to that buffer.
It's not a different terminal application, but an overlay that gives you autocomplete-style suggestions in your existing terminal emulator. I don't know all the terminal emulators it works with since it's MacOS only, but it works in both iTerm2 and the VSCode terminal, granted that you're not using a terminal multiplexer like tmux (which makes it useless for my workflow).
Interesting - I tried Fig for few days few months ago, it gave me more headaches than relief so uninstalled it. Surprised that it made into a viable product and shocked it actually got acquired.
Same experience. Tried a few times in different machines and it either froze the machine, ate up crazy memory, or other random behavior. I’m shocked this made it anywhere.
Congratulations! I've been using it for about a year or so. Hopefully their product will continue to exist... that's what their blog says now, but that's pretty much what everyone says at first.
They seem to have built a decent community, so I can see it continuing, provided the core/all code for Fig is open source. In the last 6 weeks, 57 new contributors took the time to create 62 issues and 111 new contributors took the time to comment. These are very heathy community metrics, but there hasn't been a lot of code activity, which would make sense if they were focused on the sale to AWS.
AFAICT the core of Fig isn't open source. (Why else would there be a waiting list for Linux users if they could just grab the code for Fig and build it themselves?)
What you are looking at seems to be a repository of autocompletion specs and some server to publish them.
I love this product, have contributed several times to it, and I'm a little torn. One thing I am thinking about now, is that the completion specs are MIT-licensed, and it should be possible to use them to re-implement a basic open-source version of the autocompletion product... https://github.com/withfig/autocomplete
Congratulations!? I do hope that means AWS’s ever growing cli commands will be all fig now (or will be)? It’s getting a bit long in the tooth. What about support outside of AWS? Will fig still be as awesome in iTerm2 with what-ever-is-in-%PATH%?
Any venture funded product has a sword of Damocles over its head. After acquisition the company product is much less likely to be shut down in the long run (unless it’s Google).
> Existing users will continue to be able to use Fig and will receive ongoing support. What's more, we are now making all the paid Fig Team features completely free.
can't blame them. work on what seems to be a obviously needed tool, manage it for 2.5 years, then get "not acquired" by Amazon to take it off your hands, especially since the money was coming from enterprise and larger teams to begin with.
> Within Butterfish Shell you can send a ChatGPT prompt by just starting a command with a capital letter, for example:
This is a dangerous assumption. Not all commands are lowercase. Interaction with an external service should be a deliberate, discrete action on the user's part.
I like that a lot! It would be awesome if the client running on goal mode had capabilities to request some search engine API + do some crawling. Imagine getting the info out of up to date github issues or directly from AWS docs.
I've experimented with it, the reason I haven't yet added it is that I want deployment to be seamless, and it's not trivial to ship a binary that would (without extra fuss or configuration) efficiently support Metal and CUDA, plus download the models in a graceful way. This is of course possible, but still hard, and not clear if it's the right place to spend energy. I'm curious how you think about it - is your primary desire to work offline or avoid sending data to OpenAI? Or both?
Readit News
This sounds a lot like the product is dead, and may emerge again at some point as an AWS hosted, Amazon branded product...
I'd never use a subscription + telemetry laden product like this in my core workflow, but sucks for the current users I guess.
If AWS bought it for the product, what reason would there ever be to stop the current business model entirely rather than leaving it on its current trajectory for a few months, untill said "needs" are addressed?
Just tell us if you just wanted the employees, just wanted some rights, wanted to integrate their product for your use, or just wanted to kill something off. I'm sick of reading the same new release for every acquisition knowing it's bullshit.
If it was just acquihire there probably wouldn’t have a big blog post tying it to AWS. Just negative publicity for acquirer when they could shut down quietly.
I'd assume the same thing is going to happen here.
That’s an easy one: if it’s burning money like crazy on its current hosting, but only because of rents extracted by one of their competitors, not for any inherent OpEx-intensity; and that therefore Amazon expects it to be profitable when moved to live rent-free on their hosting.
Several companies have tried to SaaSify the terminal and failed, so I suspect I am not alone here.
"We've noticed that there's a component of the computing infrastructure that isn't sending everything you do to the cloud and charging monthly rent..."
Idk though, this doesn't seem useful enough to me as an individual that I'd bother looking into it. Maybe for a team or company.
That's a euphemism. These companies "acquire technology" in the same sense that revolutionaries acquired Louis XVI's head.
Having something with autocomplete that works seamlessly with their services seems like a better idea than a plain prompt if you're going to use their shell. Hopefully they do something sensible and customer-centric with the telemetry stuff, that seemed to be the big drawback for many.
I am fully expecting that I'll have to go back to zsh-autosuggestions once the current iteration is fully wound down.
My favorite 'acquihire' story is when Oracle bought out MySQL and the creator forked it and made MariaDB.
Wiki tells me: https://en.wikipedia.org/wiki/MySQL#History
Sun Microsystems acquired MySQL AB (the company) in 2008.
Oracle acquired Sun Microsystems in 2010.
I may be becoming a dinosaur, but it's not that I'm not willing to try new things. On the contrary, after many years of being rigid about having one true development environment, I've moved away from Emacs to VS Code, and work from more heterogeneous environments instead of being 100% Mac. So these platform-specific thick client apps no longer feel like the way to go.
And this isn't coming from some FOSS ideology. I don't personally care about any of that, but the best tools are sometimes FOSS. I've got a Mac and can deal with Linux as needed.
There's a whole graveyard out there of "next-gen" terminal projects, and none of those still alive seems to be gaining dominance. I don't think writing down a spec would help this, but it would be great if whatever emerged victorious had one.
If you need reporting on the scripts your developers are running on their local machines, that probably means they have too much access to live environments. Guardrails and reporting should be in your CI/CD process and your source control.
Shells like fish or shell plug-ins already handle autocomplete nicely, so that feature isn’t that useful.
If you need env/dotfile management you probably need to spend time making a proper development environment rather than shimmying in a tool to make doing the wrong thing easier.
SSH key management is so trivially accomplished by numerous methods like configuration management or built-in solutions offered by cloud providers. Plus, most development teams almost certainly give out too much SSH access as it stands.
If you need to SSH regularly that means your application isn’t shipping logs and your local development tooling isn’t good enough.
I'm a super confused soul, and currently I use all of them:
- Emacs (for clojure / lisps / big projects).
- NeoVim (for when I want to work in the shell, smaller scripts/playground, ssh).
- VScode (js / ts / frontend / figma / etc).
I've made both my vim and emacs keybindings / features almost identical, so there is limited context switching there.
I feel very slow in VSCode, I don't even have one of my most common task optimised in VSCode yet: grepping something and quickly moving to that buffer.
I was using both before but the upcoming annual payment reminded me that its over
Deleted Comment
Have seen Brendan talk a lot about it on Twitter.
They seem to have built a decent community, so I can see it continuing, provided the core/all code for Fig is open source. In the last 6 weeks, 57 new contributors took the time to create 62 issues and 111 new contributors took the time to comment. These are very heathy community metrics, but there hasn't been a lot of code activity, which would make sense if they were focused on the sale to AWS.
Source for my insights from https://devboard.gitsense.com/withfig
Full Disclosure: This is my tool
What you are looking at seems to be a repository of autocompletion specs and some server to publish them.
Any venture funded product has a sword of Damocles over its head. After acquisition the company product is much less likely to be shut down in the long run (unless it’s Google).
This sounds like a win for users.
It's much more bare-bones than Fig but perhaps useful if you're looking for an alternative! Send me feedback!
This is a dangerous assumption. Not all commands are lowercase. Interaction with an external service should be a deliberate, discrete action on the user's part.