Phoenix ingests any opentelemetry compliant spans into the platform, but the UI is geared towards displaying spans whose attributes adhere to “openinference” naming conventions.
There are numerous open community standards for where to put llm information within otel spans but openinference predates most of em.
The readme goes into details more in depth than I will, but as a dev raised on javascript and cloud tech, I like redwood for probably the same reasons that devs raised on ruby like ruby on rails. I get to keep using all the tech i'm comfortable with but the tooling is all preconfigured and the dev experience is smooth and optimized.
My reading is that this is an all-in-one package of the current best practices stuffed into a jar. It saves you some decision-making but also ossifies your platform into a fixed point in the hype cycle. It's probably a solid platform and get you to market a bit quicker but that isn't really a bottleneck for anyone as far as I can tell.
IMO the main value of a framework like this isn't to save you time. That's what they advertise on the homepage, but I suspect that's just because it's an appealing concept to a lot of people. I think the real benefit is an increase in the chance that you build your app in a way that's sustainable. It's a pooling of experience and feedback on choices to create a whole that might not be flawlessly perfect for what you are building, but is likely worlds better than the disaster that usually occurs when you let the in house devs make too many choices.
> ossifies your platform into a fixed point in the hype cycle
Definitely true, but they have gone for some pretty established stuff here. The only ones I can see possibly being contentious in like 5 years or whatever timeline you want are Prisma and GraphQL, but even for those I'd bet on them still being sound choices for a long time. Also presumably the framework continues to develop (Rails 7 is a LOT different than Rails 1.), but that is a bit of a bet on it's adoption.
As a dev I get the appeal of something more flexible that lets you make your own choices, but I think that almost universally ends up being a curse rather than a blessing.
> I think the real benefit is an increase in the chance that you build your app in a way that's sustainable
Yes, indeed! We discuss "long-term maintainability" often in the project README and during interviews. This is one of the primary lessons Tom learned when building GitHub on Rails.
> ossifies your platform into a fixed point in the hype cycle
We are learning how to communicate these things better, but conventions, tight integration, and a "golden path" do not equate to lock-in of any kind.
If you read through other comments, you'll find a lot of examples where devs are using alternatives for the API as well as front-end libraries.
It definitely does not support 1 hand lid operation. Despite the quality not being as high as a macbook pro (I've owned several for comparison) it absolutely does not feel cheap.
In cases where you may damage the device, part of the beauty is that all of the parts are sold by Framework and come with replacement guides and any tools you may need (usually just the included screwdriver).
It’s funny how such a little thing like one-handed lid opening can add up to such a big factor in user experience. I have used MBPs for years and recently switched my personal machine to a yoga c740, which has … ok build quality but not great. One handed opening is hampered in two ways: first, the front of the machine is beveled in a way that makes it impossible to get your finger between the top and the bottom, which is infuriating. The framework has a notch for this, so it doesn’t seem to have this particular problem. The second is the obvious one, which is that the friction from the hinge is large enough to defeat gravity because the base isn’t heavy enough. I’d welcome a heavier base to overcome this, since it will also allow the case to be stiffer.
I feel like a total idiot at least a couple times a week while I’m holding my phone in one hand, saying “hang on, let me check” and doing weird contortions to get my laptop open with one hand using my belly to hold the bottom down.
Yeah but I get to build this typewriter... I derive value from the act of building and replicating the inner mechanisms of a typewriter that I could not build before. Isn't all entertainment somewhat 'wasteful' or 'purposeless' beyond enjoyment?
I'm 35 and am a keen collector of 1:18 scale diecast model cars. The sense of enjoyment for me comes from seeing their intricate details exactly replicated, from putting models from a similar age/brand/model next to one another and comparing them, from knowing that they can endure a long time without degrading and I can pass them to my son.
I fail to understand the appeal of "adult non-toy" Lego sets like these, however. It's impossible to replicate real-world stuff at any serious level of detail (the smallest brick is far too large), as mechanical devices they are flimsy (no greasing, no bearings, clumsily-weighed movements), separate sets do not stand well next to one another due to different scales and wildly varying subject matters. I do know that it is fun to build a Technic race car with steering, suspension, differentials and pistons, but such model-like stuff doesn't bring much value, IMO.
No, Legos don't and can't exactly replicate all the real world details of the things they model. It's no surprise that different hobbies have different pleasures. But surely you see that you have a niche hobby that to many (most?) people would superficially appear to not "bring much value"?
The joy of Lego is taking generic bricks and figuring out how to represent the thing you want to build out of them. With pre-designed kits like this, many people find pleasure in seeing how the designer figured out how to use piece X to represent object Y, or used a particular building technique to create a particular effect. e.g. the Lego Empire State Building uses the generic grille to great effect to render the windows of the building, and generic yellow tiles to make convincing little taxis, and some neat building tricks to create the setbacks in the tower without making it look 'lego-like' (with abrupt, brick-size shifts).
I got a sense of enjoyment from diecast model cars because I could open up the doors and pretend they transformed into airplanes. I also hated them because any steering or suspension components were generally plastic and flimsy by toy standards. I didn’t see the appeal to these things you couldn’t really play with.
People like different things for different reasons. Guitar Hero is not the same as playing a guitar, and comparing the two with the expectation that they will offer the same rewards will lead to disappointment. But, some people like both.
If you don't care about windows then you sacrifice about 20% of performance vs linux on baremetal according to benchmarks on phoronix of ubuntu bare vs ubuntu wsl.
Exactly why I made it for a 72-hour gamejam. This is still a very, very early prototype of a game, but I hope to put a little bit of time each week until it becomes awesome.
Also, thanks for the words of encouragement, they mean the world to me!
I've been running the last "nightly" release or whatever it is of 20.04 since Apr 20 on the same specs you list (3900x instead of 3950x). No problems so far on my ultrawide 3440x1400 display or with the proprietary nvidia driver. Just choose the 'safe' graphics option on boot of the live usb.
There are numerous open community standards for where to put llm information within otel spans but openinference predates most of em.
reply