I'm a Silicon Valley girlie into C++ metaprogramming and close-to-metal software. :3 Dabbling in computer graphics and nightly Clang from time to time.
I don't remember a map from the movie so I didn't know what to expect. Instantly I love this. I actually wish I could have this rather than street view in Google Maps.
Senior is the last word in use to describe myself as a developer, personally, and I think it's hard to overstate the current economical significance of this software niche. Granted that it IS a niche and nobody would recommend Mojo for, say, kernels in video games or art editors.
In the category of AI/HPC software Mojo aims to displace (which is my industry), I don't know many people who care that much about running Windows. The software I work on may only run as a Docker image or directly in a Linux/WSL. Mojo is hardly the only tool in the space (currently) lacking Windows support.
You can implement it with templates and people do things like this often. In modern C++, it's even pretty easy to make this type of trick support `constexpr`.
It's not true that Nvidia doesn't care about CUDA backwards compatibility. The invention of TileIR was motivated entirely by that problem. GPU architectures are supported by the full tool suites for at least 10 years by CUDA (that's longer than they are for graphics).
Nvidia is also a major contributor to MLIR, which they use for TileIR and ClangIR, so should a compatibility break occur it wouldn't be entirely up to Mojo to fix it.
I don't think someone mentioned yet, but the (now defunct?) company Outrun Labs built a product Revery which seems almost exactly like what Jane Street has done (to my amateur eyes). OniVim was their only end user product afaik, built on this tech.
It would only be effective if the significance of this work is clear. They certainly felt this message needed to reach people, and that it did work makes it self evident they were probably right.
Personally I hated Publix when I lived in Florida and it made me homesick. I've never had so much food come with mold in the packaging than from Publix.
I have been a ClearFlow user for over a year now. Generally I like it, but there absolutely are still common words that are hard to input consistently. The THEA cluster has given me no shortage of problems. Still a fan though.
Clang and GCC both let you use Pascal strings in C if you would like (with `\p`). But Pascal strings aren't that useful today because the maximum length is too short.
I feel like I've never gotten a compelling explanation for why Nastaliq is hard/unavailable. I'm not an expert on abjads, but it doesn't look harder to render then Naskh (and it self-evidently is possible since the fonts exist). Does anyone here know why they make it difficult? Urdu is much less obscure than, say, Sharada or other languages with Unicode support. I think Punjabi is also often written in Nastaliq when it's not in Gurmukhi or Roman.
I've got really comfy `just` scripts for generating Clang "intermediaries" in my CMake project. I can generate `.ii` files which get formatted and edited in a manner making them directly recompilable, along with `.ll`, `.bc`, and `.s` files. All the above are per-translation unit or post-LTO and I can constrain the output to specific functions or files, and the LLVM bitcode can take optimization passes or optimization levels to very easily introspect how my work in this codebase optimizes.
I've also got a clang-repl wrapper for this codebase that is very easy to use and makes interactive programming much easier for me.
The point here is that these examples need to be updated because @cImport, the syntactic feature of Zig, is being deprecated and removed. That there exists an alternative isn't relevant.
These demos honestly look pretty good to me. But it is objectively true that this and similar technologies are used at huge scale by every leading autonomous vehicle manufacturer, so we can inductively reason that it _is_ good enough for that use-case. I don't work on Cosmos, but I am currently working on a superficially similar non-open technology at Nvidia used by many of these leaders which, in my opinion, produces similar quality. Some of the open research for it is here:
Does anyone else miss KWrite? I had it configured as a very slightly more advanced Notepad.exe clone. I really enjoyed opening it for quickly writing out thoughts that crowd my ADHD brain, and I feel like the full Kate takes much longer to open up and looks much much heavier and emotionally oppressive for what I want to do.
Fwiw, nobody has ever suggested to me that I employ token compression in my daily workflow. I don't pay full attention in all the AI workflow demos I'm supposed to attend, but I don't recall that even being discussed. Is this an Nvidia blog or tweet you're referencing? I'm actually interested to see what they have to say.