Hacker News Reader

Comment by mitthrowaway2 | Hacker News Reader

I didn't know that there was a FreeCAD Day! But I'll take this opportunity to just give thanks to the FreeCAD developers for creating and sustaining an amazing program. We all know it's not perfect, but it's already incredibly powerful and I'm now using it professionally in place of SolidWorks. Not only do I save a ton of money, I feel way more in control of my work, and have ownership of my hard-earned skills. I can write my own Python routines to automate parametric modelling and use FreeCAD as a back-end for doing geometric calculations. It's fantastic and I'm grateful for it.

Thank you.

View on HN

I do not want to sound negative towards something that is a gift but when I tried to use it in place of SW and it is so far behind from a practical point of view.

Have you used Catia before? I used to think Catia was the dumbest CAD package ever. After being forced to use it for about a year, I honestly get why they’re made that way now. And dare I say, Catia is actually a pretty good CAD package. (I’ve lived and breathed solidworks for about a decade, used and trained in Creo, Inventor, and now Catia). I think FreeCAD is modeled (pun intended) to be more like Catia hence why it feels so far behind.

No Catia. Here is my story. While I mostly do software at some point I needed to design some piece of equipment from scratch. At the time I happened to have access to Solidworks and without any prior experience in mechanical CAD I was able to do it in about a month. Parts, assembly, subassemblies, regular CNC parts, sheet metal, welds, fastening. Stellar. Despite absolutely insane power the tool is super easy to use for complete noob. I tried later on to do something like this in FreeCad - never fucking again.

Yes, SolidWorks is easy to pick up and use with no training. FreeCAD is not; it requires dedicated time to learn, following tutorials, and at least a few throw-away projects to gain proficiency. There are a lot of easy mistakes that can sabotage progress, and multiple ways of doing things that are different under the hood, where those differences aren't obvious, including some ways that are bad and wrong. But there's a lot it can do, and it gets better all the time.

My goal was to produce the design, not to master design tool. The less time I spent the better is my ROI. I have neither time nor desire to dedicate myself to be a master of particular tool.

>>> While I mostly do software at some point I needed to design some piece of equipment from scratch. At the time I happened to have access to Solidworks and without any prior experience in mechanical CAD I was able to do it in about a month. Parts, assembly, subassemblies, regular CNC parts, sheet metal, welds, fastening. Stellar. Despite absolutely insane power the tool is super easy to use for complete noob. I tried later on to do something like this in FreeCad - never fucking again.

>> Yes, SolidWorks is easy to pick up and use with no training. FreeCAD is not; it requires dedicated time to learn, following tutorials, and at least a few throw-away projects to gain proficiency. There are a lot of easy mistakes that can sabotage progress, and multiple ways of doing things that are different under the hood, where those differences aren't obvious, including some ways that are bad and wrong. But there's a lot it can do, and it gets better all the time.

> My goal was to produce the design, not to master design tool. The less time I spent the better is my ROI. I have neither time nor desire to dedicate myself to be a master of particular tool.

This is actually how a lot of the "Niche language $FOO is a superior language" arguments go.

It's also a counterpoint to "but it's doing a complicated job so the UI has to be complicated" nonsense defence that gets brought up in FreeCAD's defence a lot.

> It's also a counterpoint to "but it's doing a complicated job so the UI has to be complicated" nonsense defence that gets brought up in FreeCAD's defence a lot.

That defence is used for $SUPERIOR_LANGUAGE too. Look at the large numbers of devs who feel that the extra cognitive load needed to learn (for example) monads is worth it.

Or the large number of posts explaining Rusts syntactical mess as "It's doing a complicated job so the interface has to be complicated."

Hell, nevermind languages, that argument gets thrown across the entire tech stack - the submission yesterday about complexity had a number of posters who felt that there's no additional complexity in using Kubernetes even for an application that never needs to scale horizontally!

Is it any surprise then that we see the same argument for different tech, even though it's a clearly invalid argument 99% of the time.

No, it's not surprising. It's a failure of abstraction design. But that's ok, designing a good abstraction that correctly hides things that can be hidden, in a composable, comprehensible way, is a really, really hard job. It's ok to fail at that.

What's not ok is denying that failure and gaslighting users into self-blame while preventing trying something better.

> What's not ok is denying that failure and gaslighting users into self-blame while preventing trying something better.

Forgive me, I wasn't attempting to imply that the practice is okay.

I was just pointing out that the practice is normal, across the entire tech stack.

It is common for people to reject high-cognitive load stuff, and at this point it is common to have others jump in with "they're dinosaurs resistant to change", "Every $TECH has a learning curve" or similar stuff.

No worries, we're vigorously agreeing :-)

People talk shit about CATIA all the time, and I never understand why. There are not many things that can do what CATIA does. Big assemblies, things with a huge number of moving parts, are all CATIA when they're not NX, because you actually can open them and move them around in it. Formula 1? CATIA. Large Hadron Collider? CATIA. Airbus? Guess what?

CATIA is rubbish. But everything else is way worse.

I was not talking about CATIA as I have never used it, I did not even know it exists. All my musings were about SW vs FreeCAD.

I did not need to assemble the airplane / LHC so this particular benefit you are describing is / was completely irrelevant to me.

Whoops. My apologies. Thought when you said "No catia", you meant "Never again" or something along those lines. Didn't realise you'd never used it.

(Not GP) I've never used Catia and I'm not in mech eng, but to me FreeCAD seems like an MVP Fusion360 clone inexplicably made with a '90s GUI toolkit.

I'm glad it exists, and appreciate the herculean task that it is, it's just a shame, and it really is clunky/unintuitive compared to F360. Personally I think some focus on UI & polish would go a long way, drive more adoption, which is presumably somewhere on the long road to more support in development and funding.

I get the impression it's probably made by and for a few though, scratching an itch and not too bothered about adoption. Even among hobbyists it seems to be a fairly uncommon choice - pick a Youtube channel and it's usually Fusion360 I find (and as far as I've seen not sponsored, unlike Altium Designer say which is an even weirder (more limited/expensive) choice for a hobbyist).

As a hobbyist I can't imagine using anything but OpenSCAD. I know it's not comparable in capabilities and has its quirks, but it feels like Python for CAD - I get stuff done quick. It's so simple the cheatsheet covering 100% of commands takes up one page, so I don't need to waste time looking up details. And as it's all code it spares me from clicking through all kinds of menus.

I do miss FreeCAD's chamfer & fillet tools, though.

Where FreeCAD ought to be able to take over is in the things that are a bit more complex that OpenSCAD just doesn't have the tooling for. Multiple-part assemblies is one of them that immediately comes to mind, now that there's some reasonable FreeCAD functionality for it.

Fillets.

>Python for CAD