Autodesk criticised by architects

> It's one of the many reasons people stick with their 75% or 85% good enough present solution.

100% this. That's one thing many people don't realise until they run a business - there are so many things which MUST be done, all the time, just to keep the lights on. When you do get a rare gap in those, you're better off developing and improving your core product than indulging yourself by spending resources on some unrelated nice-to-have.

Version management for more structured data is hard. In a previous job we did extensive work on it in the GIS domain, and even having everything backed by a long transaction version controlled data store, resolving conflicts was not always easy. We did a lot of work on ways to automatically resolve conflicts, and it was only worthwhile for large customers who could set up good rules for their workflows (e.g. if you moved a pole in as part of a resurveying workflow then that position should take preference even if other changes happened to the pole).

We can handle this sort of thing in software because we can try compiling something, see it has errors, and fix the obvious syntax problems that have been introduced by a merge or rebase. We have tools that make it easy view and copy data from a previous version, and all of this makes things easier. For more structured data those low level conflicts need to be handled much better so that the user can concentrate on the higher level aspects. If they need to restore sub structures deleted in a merge then they need to be able to see those, so our version control starts to bleed right through the tool chain.

Is there anything a non-professional can use for the following four usecases (separate software is ok):

1) Reinforced concrete structure design, with support for non-trivial surfaces (e.g. splines/Bezier curves/surfaces).

2) Non-finely-meshed (in comparison to local volume thickness) truss structures, ideally not just classic steel beams + welded/riveted joints,

3) but also aluminium (considering fatigue from non-stationary loads and optimizing for the lifespan/weight/cost pareto frontier (cost would be just a simplistic metric)).

4) The holy grail, a 3d-printed core (technically a large-pore-size, low-density, open-pore foam) that is then covered by fibers via a robot that passes the spool around the core/preform (filament winding). After hardening the matrix that bonds the fibers together, the 3d-printed core could optionally be removed by melting and/or solvent washing. This kind of structure is potentially extremely stiff due to combining the truss structure with cylindrical beams and a lightweight fiber-reinforced polymer material.

The issue is just the insane non-triviality in designing such a structure, because there are restrictions to holes and fiber angle shifts due to the winding process, where the fiber bundle has to be accurately woven around the mesh in the designed way. This, combined with the directionality of the fibers makes it necessary to consider much more than normal 3d-printable topology optimization[0].

A concrete-ish example would be a multi-monitor + keyboard/mouse support structure to enable a low-fatigue position where the user leans back about 30~60 degrees to allow the neck to be supported and the user's practical FOV (w.r.t. feasibly eye/head movement) is covered by screen area and the arms can rest on supports. The constraint that makes a traditional steel truss undesirable would be to have it mobile (thus lightweight), yet stiff enough to handle ~5 m/s² without impairing usability. A practical case would be to overlay travel with getting work done, by mounting this in the back of a van (including straps/seatbelts as required) and being driven.

A different practical example could be amateur airplane design, e.g. devising an autonomous solar glider, where some stiffness is required to not break the solar cells laminated into the wing surface (thinned wafers are lighter and can conform to the wing's curvature, but will then dislike the wing flexing along the span), but weight is extremely important. Mounting the heavy things (motors, some backup batteries, potential payload) and handling control surface actuation forces won't easily work with a simple fiber-reinforced hollow wing, as a lack of dedicated mounting points (with their own fibers distributing the forces into the outer structural surface) would require a far heavier construction.

It's also difficult to consider aerodynamic forces when computing pre-distortion of the unloaded structure, so that design loads make it take on the desired shape, if there are also ad-hoc mounting brackets glued to the inside surface (which would be necessary if they're not designed into the primary structure).

Sorry for the slightly-OT examples.

[0]: Project video "on using the free version of Fusion360 for shelf brackets printed in PLA on a Prusa i3 MK3": https://www.youtube.com/watch?v=3smr5CEdksc - Thomas Sanladerer: Making STRONG shelves with Topology Optimization

> Small company rises to produce quality software, it sells, responds to bugs, listens to users, adds features etc.

I would say:

Small company invests time and money into creating a decent enough product. They listen to customers to build their roadmap and respond to most bugs in a reasonable time. Small company could grow to a medium or large and still hold true to their roots.

Large company buys a mediocre to decent company instead of doing R&D. New company is integrated into large companies suite of products.

Large company fixes only enough to keep the product running or shuts it down completely. Support loses its personal touch and is transferred to the cheapest FTE location. Large company’s corporate politics causes infighting between departments with similar products.

Large company merges with equally large company to reduce competition. Execs cash out and leave ASAP. That’s where I work now.

Using a hot key 100 times a day for 10 years sounds like exactly the kind of thing software engineers SHOULD understand. I heard a lot of anger from Vim users when Apple messed with the escape key.