David Raffauf

O que era apenas uma lenda urbana contada em fóruns obscuros de BBS e mesas de bar de engenheiros aposentados acaba de ser confirmado. O Retropolis teve acesso exclusivo a imagens e documentos técnicos do Gradiente Expert Turbo (Modelo G-90), o computador híbrido que prometia unificar a informática nacional com a revolução dos 16-bits da SEGA. O Elo Perdido entre o MSX e o Mega Drive Em meados de 1989, enquanto a Gradiente colhia os frutos do sucesso do Phantom System, um projeto secreto corria nos laboratórios da Zona Franca de Manaus. O objetivo era audacioso: criar um computador pessoal que, além de ser um MSX 2 pleno, possuísse um barramento de expansão capaz de ler nativamente os cartuchos do Sega Genesis (o nosso Mega Drive). O protótipo encontrado, com o número de série 0001-XP, traz o gabinete clássico do Expert em um tom de cinza chumbo fosco, com um detalhe que faz qualquer colecionador tremer: um segundo slot superior, ligeiramente mais largo, com a inscrição “High Grade Multi-Link”. Arquitetura Dual: O Coração da Máquina Segundo o manual técnico recuperado, o Expert Turbo não utilizava emulação (tecnologia inexistente na época). A máquina era equipada com: Processador Zilog Z80A (3.58 MHz): Para compatibilidade total com o padrão MSX. Coprocessador Motorola 68000 (7.67 MHz): O “cérebro” do Mega Drive, que permanecia em estado de hibernação até que um cartucho de 16-bits fosse inserido. Chip de Som Yamaha YM2149: Capaz de mixar o som polifônico do MSX com o sintetizador FM da Sega, criando trilhas sonoras inéditas. “Nós tínhamos o protótipo funcional. O Sonic rodava perfeitamente enquanto o WordStar estava aberto em segundo plano,” afirma o Eng. Alberto Santo Antonio, que trabalhou no projeto. Por que nunca chegou às lojas? O motivo do cancelamento foi puramente comercial. A Tectoy, que já costurava o licenciamento exclusivo com a SEGA no Brasil, teria entrado com uma liminar impedindo a Gradiente de lançar qualquer hardware que utilizasse o chip 68000 com a marca japonesa. Sem o apoio da SEGA, a Gradiente decidiu “capar” o projeto, transformando o que seria o Expert Turbo no que conhecemos hoje como o Expert Plus e o DD Plus, removendo o suporte aos 16-bits. Agradecemos aos nossos “informantes” UltraMan, UltraSeven e Godzilla por essas revelações. Parece que eles deram uma pausa em salvar o mundo para desenterrar os segredos da Gradiente e da Sega!

@PapaJohns, your ordering system's checkout has been broken for at least 30 minutes. I had to order by phone and pay about $14 (35%% off) vs about $9 each for three larges (online deal) since stores don't have access to the same deals.

Microsoft doesn't want you to know this

@bitmap_books, are you making more of the SNES RPGs Collector's Edition? You're out of stock on pre-orders.

Anyone know of examples of actual fractional arithmetic being used in software to reduce accumulated error? e.g. storing both numerator and denominator of something like 1/3 seperately, and doing algebraic multiplication rather than reducing to a single float.

Does Django/Python have any good query analyzer and optimizer tools, similar to the great active_record_doctor? I always assumed ARD relied on EXPLAIN, but when I browsed the code, it was all Rails introspection.

This sounds like a @moldilox jam. youtu.be/uVVb8hOyQnQ

I've spent the last 2 weeks learning about Nginx and Assembly in order to make a very small HTTP server. It's currently around 5.3 KiB. • no dependencies/libc/bloat • multiple processes managed via epoll • single arena allocation and it is very, very, very fast.

@InsaneMegaCD It's a late-90s retro variation of the tacticool aesthetic.

Am I the only one who finds this kind of modern, mass-marketed "gamer aesthetic" incredibly cringe?

If you have multiple monitors, it's way more than 20 commands.

DX12 is 10 years old now. It's not modern anymore. Most of the complexity of DX12 and Vulkan APIs is simply not needed anymore. We don't need to support 13 year old Intel iGPUs and 12 year old mobile chips anymore. Metal 3.0 binding model is much more clean. Just GPU pointers.

The JS ecosystem has 3-5 serious choices for every possible layer of the dev stack. It's dizzying just starting a project.

@bitmapbooks @bitmap_books

I have a mini @bitmapbooks library.

Vibe coding at its finest

I was tired of manually switching around inputs and daisy chaining displays and dongles, so I purchased a highly rated KVM switch. It was under $100, the extra cables were another $40, and it all works like a charm. I should have done this a decade ago!

This also assumes that rising development costs and profit from increased market sizes have remained at a constant ratio, which seems very unlikely.

First ever documented instance of Vibe Coding was the StackSort by xkcd

When things go from 3D to 4D in the graphics pipeline:

I should clarify that learning OO and Lambda Calculus were eventually useful, but probably too abstract as a starting point. In comparison, my Data Structures and Algorithms, and Assembly / Discrete Matg classes have been very useful all along.

My Computer Science program actively derailed me from learning programming. We learned lambda calculus, OO, and RDMS poorly. I would have been better served studying C and making games. I studied web dev independently from 1995-2003 and that's what got me jobs.