This seems a hard thing for many if not most devs but even in the most hardcore software company most employees are not devs. If anything the smartest people in your org are not the devs (unless you’re FAAnG). Any tool that puts more power into the non devs is a great tool. Any person managing a company that needs to string apis together and perform high touch service would need a tool like this. Now why this tool and not any other RPA is a different question.
Look at the examples though. See the second one. In the animation, it shows the dot distributing the different values within a response, as though it's some parallel task. Outside of that it's serializing an array on time for no reason I can tell. That's like the Jump to Conclusions Mat in Office Space...
Postman somehow managed to make a slow, bloated, piece-of-crap api-tester that's somehow wildly popular. Extrapolating from that, I expect this to be incredibly successful regardless of how good it is.
Postman was a huge success because it was lightweight and had just the right amount of features. When they took the success to their head and started taking VC investment, building a API testing client was no longer enough and that's when they started putting everything and their mother into it, and they lost their moat.
To date, Postman has accepted 433,000,000 USD in funding. For a API testing tool that slowly but surely expanded in scope.
$433mm sounds high but then again, Calendly's raised $350mm and 1Password's raised $920mm, so maybe that's just where things are these days. Without a view of Postman's internal roadmap, we can't say for certain what their world domination plans are, and why there's been that investment (we don't know what the valuation the $433's been bought in at either). Postman Flows is pretty clearly the low/no-code future they're betting on, and given a more advanced, developer friendly interface to a Zapier-like product (a fully managed low code PaaS offering), the bridge between a low-coder+platform fees, vs sr fullstack dev+cheaper platform, theres a comparison of total costs to be done, with lower pay for a low coder + higher platform fees ending up being less than higher pay for a sr full stack dev + lower platform fees.
I use Postman every single day and have very similar feelings towards it as you. Simple things that would make it exponentially more friendly to use just aren’t there, and I can’t believe they haven’t been added in the years it has existed. It updates itself occasionally but never gets any better!
I used to be a fan of Postman (in the early beginnings of the tool), but as it added features I didn't use and became slower and slower, I got tired of it.
Eventually, I figured I should try something true and tested, so I went for cURL. It takes a bit of practice and memory to learn how to do even the basics, but once it's there, it's a rock solid tool, with support for almost everything you could think of.
I used to use Postman exclusively, but like you started trying curl.
And then I realized curl was absolutely brilliant for my common uses because I’d write simple scripts that would mean all I needed to pass in were the arguments that changed.
I took it another step further and had my script lookup a folder that had a whole bunch of different types of requests that I could then trigger with a single command.
Finally, I created a version that plugged into fzf and made it super easy for me to run a whole bunch of common scripts trivially from the command line.
The great thing about command line based programs is the incredible ease with which you can create repeatable series of commands that can be run both interactively or programmatically, and the UX can be tailored to suit the specific needs of what you’re trying to do.
I’ve recently started using Thunder Client in VS Code and it seems … good? Sensible, lightweight. And the IDE feels like the right place for this. I haven’t used Postman that much so couldn’t comment on the comparison.
As do I. It's lightweight and responsive and its UI is no-nonsense and doesn't try to be branding or chase trends. It's just another tool in the toolbox and acts accordingly.
well it can eat around 300mb of ram, but its still very responsive and fast, so I don't mind that, but that's not lightweight, unless postman tops that and eats much more ram, then sure, rapidapi is lightweight
There were several other alternatives but it has been a year since I looked at it. I do recall that none of them use the same export/import format hence I never bothered moving.
It’s really frustrating. I used to “live” in postman but it became a chunky mess… and I haven’t found a good home for quick api testing, browsing… yet.
I don't know why but I'm always tickled when someone comes up with yet another data-oriented / flow diagram programming language like NI LabView. And my first thought is always "this will break down exactly like LabView and for the same reasons." Someone will inevitably come up with a use case that is just a little too complicated but without any mechanism to go from data flow to program code there are going to be some really painful migrations.
There wasn’t a single day i used labview that I did not wish it was a simpler CLI or text file based program. Heck, even an imperative text based programming language would have been fine.
And Labview was one of the best Visual programming tools I’ve used and possibly in existence.
I love the idea of this. Reminds me of Yahoo Pipes, which was years ahead of its time.
But I'm not so sure they're going to get the adoption they're after.
Even the developers I know who are enthusiastic about Postman for its core offering haven't really responded to Postman's push to move more of the development cycle into the application, and I can't imagine this will be the thing to pull them in.
It depresses me that SaaS is called “software” in the same way that software you can run is software.
Plugging together APIs to services isn’t the same thing as creating software people can actually run.
Pragmatically your service Rube Goldberg machine breaks as soon as anything it’s made of makes a breaking change or goes down.
Then there’s the fact that anything built this way has zero privacy, a massive security surface area that you can’t even know let alone control, and eternal rent that must be paid for every piece that can be raised at any time.
What a dystopia we created. I say we because we all did it. We had to make software “free” and killed commercial software business models, but the only thing we accomplished was to push it all to the cloud and to a model that is substantially more closed than closed source commercial ever was.
Lesson: you get what you incentivize. Always, always, always ask “what is being incentivized here?”
I would love to see a new term to replace "software development" that describes the sorts of task that Postman (Flows) are appropriate for. "Application development" is a bit of a throwback, but it's a good start.
This stuff has nothing to do with the majority of software development if measured by task type, even if it has everything to do with the majority of "software development" if narrowed down to browser/web-centric data presentation and editing domains.
This has nothing to offer embedded systems, creative software, operating systems, high performance computing, LLMs and so much more.
Sure, it seems super reasonable. That's why you pay for things like Python, GCC, Go, etc. It's also why Chrome dominates the browser market -- they can innovate because of the subscription fees they charge.
I’m trying to figure why they consider this the next generation of software dev when there are quite a few orcastration tools similar to this that have been out for a while and have a community behind them. n8n or node-red are just two examples that come to mind.
Their demo looks nice in terms of the ‘flow’ building UX. I’d be willing to bet they’re using React Flow[0] as the primary rendering engine of this tool.
APIs aren't a step forward, they're a step back. They're a terrible abstraction. Their interfaces are versioned and often not backwards compatible, data exchange is often poorly documented using no common conventions, the underlying HTTP interfaces vary greatly, there's no universal way to read the docs, and there is no single standard or convention for them.
Compare to Unix. Single way to run something, one simple return status, one basic input stream, two basic output streams, standard key=value pairs that can be inherited and passed on, generic arguments, there's usually a man page that fully explains how to use it, and they can be infinitely chained.
APIs require way too much "integration" and provide too little utility, and zero composeability. They're extra work with less usefulness. But, hey, "it's the web", and we're all too chickenshit to buck any trend that starts. We're slaves to whatever shitty paradigm takes over on "the web".
The web killed computer science innovation. We're just slowly and poorly rebuilding an operating system on top of a browser. But with more complexity and less utility.
I think not. It seems like Yahoo! Pipes. It's cool but I don't see most devs using it. And FWIW Scratch is way cooler.
It could still be nice but that's exaggerating it a whole lot, I think.
Pipedream could be much further along. https://pipedream.com/