4 ms·
Somethign I haven't seen mentioned here is that the technology used to design this thing is waaaaaaaaaaaay old. I know for a fact it was designed on an IBM mai
by mgarfias 6y ago
Somethign I haven't seen mentioned here is that the technology used to design this thing is waaaaaaaaaaaay old. I know for a fact it was designed on an IBM mainframe (dad ran the one that the YF-22 was designed on). I can't remember what CAD package they used there, but its definitely conceivable that the software doesn't even exist anymore. How would you go about building a new one if you can't read the CAD files, if you have them.
Hell, the building in hawthorne where dad worked isn't even owned by northrop anymore. Its spacex.
- folli 6y agoReverse engineering the files night be easier than reverse engineering the airplane.
- curiousllama 6y agoSays the guy not constantly surrounded by aerospace engineers :) What are the odds someone said the phrase "those parts? I could mock up copies in a weekend!"
- adolph 6y agoYour story reminds me of what I read about the McLaren F1, that it had a particular 90s Compaq that had to be kept running in order to service them. As it turns out, this is recently no longer true. He also has "the laptop"—a mid-1990s Compaq running custom DOS software created by McLaren when the F1 was new. This gray brick was, for many years, required to access the car's engine control and body control modules. Today, McLaren uses a modern Windows computer running a software emulator for day-to-day computer maintenance. Hines keeps the vintage Compaq around just in case. https://www.roadandtrack.com/car-culture/classic-cars/a14453949/the-only-mclaren-f1-technician-in-north-america/ https://www.roadandtrack.com/car-culture/classic-cars/a14453...
- triggercut 6y agoGood callout. However, it might not be all that bad. CAD file formats are known to last for decades to maintain backwards compatibility, since the packages are expensive and upgrade paths of large companies slow. CAD files (at least planar modeling as opposed to parametric or solid modelling) tend to have pretty straight forward data structures for geometry. Find the point or vertex struct and you're halfway there. The complexity is association and metadata which were nowhere near as bloated then as they are now. As someone who works regularly for large asset owners, asset life-cycle information (digital and analogue) and the proper management of that information is often ignored because organizations pass that risk to the people working with the asset maintaining their own working knowledge to get their job done. We've had international Building Information Modeling standards since the turn of the century but there's often no business case or standard enterprise process to make leverage technology for asset management. AI and ML are starting to appear more and more in the operations space, however many pilot projects have to start with digitizing the asset in the first place, which becomes a cost impediment for many getting off the ground. It really frustrates me as there are great things that can already be done today but the initial step financially usually scares off anyone who has to know 5 years in advance what their operating budget is.