Assembly3, A2plus, Assembly4? Get united!

Discussion about the development of the Assembly workbench.
User avatar
Zolko
Posts: 649
Joined: Mon Dec 17, 2018 10:02 am

Re: Assembly3, A2plus, Assembly4? Get united!

Post by Zolko » Mon Jan 13, 2020 7:56 am

ColdAK wrote:
Sun Jan 12, 2020 4:12 am
I think the biggest worry and concern is wondering which one or ones will be around long term, not that there are multiple options ... I use freecad for business and I actually would like the peace of mind that the method I use will still be supported years from now
yes, I agree, I have been saying this same thing often (for example here). But this being software, developers prefer to work on features because it's more exciting than working on code stability/maintenance. And this is very unfortunate as long-term user-data accessibility could be FreeCAD's strong selling point, because that is something where commercial CAD vendors really suck. And even more so with the current tendency of SAAS (Software As A Service) where the software is "rented" and not "owned".

ColdAK wrote:
Sun Jan 12, 2020 4:12 am
For example, drawings I make today need to be valid or accessible with minimum tweaking to bring it up to date 10 years from now
well, this is the Assembly forum, for drawings you should address your queries to the TechDraw forum :-) That being said, my concern is to be able to access my model at-all ! I always export my finished drawings to PDF so they're accessible for ever, and if needed I can re-implement them in another drawing software. If your business is dependent upon these drawings then you should have some product lifetime management system anyway, you can't rely on the CAD software for that (even though some pretend to do that, I know). But if I can't open my assembly, then I'm really toasted. Which is what happens if you use a commercial CAD software, and it's called vendor lock-in: you won't have this with FreeCAD.

On the other hand, what you'll have is that some feature depends on a single developer, and when he's not around anymore that feature might disappear: see Assembly2 (not A2+). That's why Assembly4 only uses stock FreeCAD things without any magic, so you'll be able to access your data as long as FreeCAD exists. You don't even need the Assembly4 workbench to be installed to open Assembly4 models !

Hum, now that I actually get to tell the story, I'll be even more careful about this, thank-you for your comment.
Last edited by Zolko on Mon Jan 13, 2020 11:52 am, edited 1 time in total.
try the Assembly4 workbench for FreCAD v0.19
install with Tools > Addon Manager > Assembly4 — tutorials here and here
chrisb
Posts: 21445
Joined: Tue Mar 17, 2015 9:14 am

Re: Assembly3, A2plus, Assembly4? Get united!

Post by chrisb » Mon Jan 13, 2020 8:24 am

Zolko wrote:
Mon Jan 13, 2020 7:56 am
That's why Assembly4 only uses stock FreeCAD things without any magic, so you'll be able to access your data as long as FreeCAD exists. You don't even need the Assembly4 workbench to be installed to open Assembly4 models !
That's indeed a big advantage.
When asking about a single Assembly workbench I was hoping for something finally belonging to the FreeCAD core system. That would mean two things: It has more taste of reliability as it is most probably around as long as FreeCAD is; and it will be respected by changes in other workbenches.
User avatar
Petert
Posts: 29
Joined: Tue Dec 01, 2015 9:27 pm

Re: Assembly3, A2plus, Assembly4? Get united!

Post by Petert » Mon Jan 13, 2020 10:10 am

vocx wrote:
Mon Jan 13, 2020 5:53 am

Sure. It's free software, so the code is there for anybody to maintain and keep alive forever.
And that is a giantic bonus. I kept some older versions just to be able to use my designs in the future.
vocx wrote:
Mon Jan 13, 2020 5:53 am
The problem, as you know is that you get what you pay for. You don't pay anybody for the software, then you can't expect much in return. If this is a deal breaker for you, then you shouldn't be using this system in the first place. Pay your dues to FreeCAD, sponsor a developer, and you can see this situation improving rapidly.
You are missing the point here and being very obnoxious about it too.
We are trying to make a point about some future direction for the assembly system. Since the assembly systems are not compatible with each other and it is very unlikely that there will be multiple systems in the future, the solution for being able to continue the assembly wb lies in a single system. That way we can keep our investments in time and effort save.

It takes considerable time to learn the wb's, each has his/her own quirks and it would be really bad for FC if we had to learn a new approach just because there is no clear direction.
I am a refugee from Fusion 360, have been using Fusion for almost 4 years now and after the latest debacle with the licensing stuff I had enough of AutoDesk.
Eager to switch over to FreeCAD and spread the word.
vocx
Posts: 2841
Joined: Thu Oct 18, 2018 9:18 pm

Re: Assembly3, A2plus, Assembly4? Get united!

Post by vocx » Tue Jan 14, 2020 12:37 am

Petert wrote:
Mon Jan 13, 2020 10:10 am
... it is very unlikely that there will be multiple systems in the future, the solution for being able to continue the assembly wb lies in a single system. ...
You are the one missing the point. Why should there be a single system? There is no basis for this statement. I'm sure there should be a single Linux distribution, right? And that one should be Ubuntu, right? Then, why do we have all these Linux varieties? Why don't they just create a single one? Because that's how free software is. People decide what works for them or not freely.

That you take a long time to learn something is not a going to go away, that's just how life is. Complex systems take time to learn. If we never innovated on the first solution that was made available we would not have progress. There is no point at which you would say, "this is finished, no more is necessary". I mean, you could arbitrarily say that, but that won't stop progress in realistic terms.
To support the documentation effort, and code development, your donation is appreciated: paypal.
User avatar
Zolko
Posts: 649
Joined: Mon Dec 17, 2018 10:02 am

Re: Assembly3, A2plus, Assembly4? Get united!

Post by Zolko » Fri Jan 17, 2020 12:14 pm

Petert wrote:
Mon Jan 13, 2020 10:10 am
the solution for being able to continue the assembly wb lies in a single system. That way we can keep our investments in time and effort save.
That depends on what you call "system". What really matters I think is the data structure: that a user can re-ope designs made in old FreeCAD versions, and still re-use them in new versions. Else, nobody will ever use FreeCAD for serious stuff.

The tools, on the other hand, can change and evolve. Should, actually, evolve, even though it will need some new learning. But the data must be preserved.

So my take on this subject is to document the data structure used in Assembly4 and see if it's possible to make it converge with other assembly workbenches. On the other hand, I'm also quite sure that there is not 1 and only 1 best solver for all problems, and we should not try to force a single solver on every user, as long as all solvers use compatible user data structures, so that the user can keep his investment on the long run.
try the Assembly4 workbench for FreCAD v0.19
install with Tools > Addon Manager > Assembly4 — tutorials here and here
fcaduser
Posts: 240
Joined: Wed Feb 19, 2014 6:53 pm
Location: Near Paris, France

Re: Assembly3, A2plus, Assembly4? Get united!

Post by fcaduser » Tue Jan 28, 2020 10:21 am

Couldn't the constraints solver at least be shared using a common module ? Or are the data structures so different there is nothing meaningful to do ?
frabice
Posts: 28
Joined: Mon Jun 09, 2014 6:21 pm
Location: Clermont-Ferrand, France

Re: Assembly3, A2plus, Assembly4? Get united!

Post by frabice » Sat Feb 22, 2020 12:39 pm

Hello,

I am a mechanical engineer and have a good knowledge of PTC Creo (use at work) and a small knowledge of Solidworks. As I wanted to get a better overview of Freecad and assembly workbenches, I decided to model a small motor from this website : http://www.vapeuretmodelesavapeur.com/telechargements4/
I chose vertical beginner 1.

I learnt a lot on part design workbench, discovered sheet metal workbench. But my goal was to test every assembly workbench to try to catch the differences between each one and see which one suits best to me. I thought I would share my feelings on the forum so that it can help developers and/or other users. I do not mean to be hard to developers and I appreciate the effort it takes to get where Freecad is. It is just a little experiment for me to choose the workbench I am the more comfortable with. You will find below what I liked and what I liked less for each workbench. I do not intend to get a response on each point from developers but just give my impressions. They can be used or not, it is up to everyone. I could also elaborate if needed.

A2+:
Capture d’écran 2020-02-22 à 13.22.06.png
Capture d’écran 2020-02-22 à 13.22.06.png (43.62 KiB) Viewed 198 times
+ :
- There is a button to import part
- Automatic placement when constraint is given
- Similar to commercial CAD software : learning curve is fast
- Only possible constraints with current selection are active in toolbar

-:
- Cannot access to part datum object for constraining
- Unable to know which constraints are conflicting
- Cannot rotate part when moving under constraint (I did not get it to work)
- Miss plane/cylinder tangent constraint
- Constraining with 2 coincident axis to lock rotation does not seem to work

assembly 3
Capture d’écran 2020-02-22 à 13.22.22.png
Capture d’écran 2020-02-22 à 13.22.22.png (34.68 KiB) Viewed 198 times
+:
- Possible to access all parts geometry and datum to constraint

-:
- Need to move part before constraining
- Constraints not clear about what they do, it did not feel natural to me
- No import part button : drag and drop is not always easy with a big tree
- Need to prepare all interfaces object before constraining if you want to ensure reusability
- You have to open, save, reopen to start constraining
- All constraints are always available active no matter current selection
- Miss plane/cylinder tangent constraint
- No « reverse direction » option on relevant constraints
- Lot of clicks to do something
- Need to create assembly files before starting to work
- Solver sometimes tells you inconsistent constraint with what you think are consistent constraints
- Solver fails after moving a part even if it was OK before moving

assembly 4
Capture d’écran 2020-02-22 à 13.21.54.png
Capture d’écran 2020-02-22 à 13.21.54.png (42.28 KiB) Viewed 198 times
+:
- Good resilience to toponaming problems
- Once LCS are placed, constraining is easy

-:
- You have to create a model, save, close, reopen before starting working
- You have to create each LCS. I would prefer automatic creation when constraining by geometry’s object
- You have to edit placement manually. A popup window at LCS creation would be nice
- LCS placement is sometimes not natural compared to what you would do in real life for assembly
- No « move under constraint » feature, only animation
- No access of subassemblies’ LCS (is it possible to work with subassemblies at all?)
- Cannot create 2 constraints on the same part : you have to use master sketch. Not my habitual workflow so not natural to me, but it works. To me, part offsetting from master sketch is more complicated than putting a plane coincident constraint.

I did not succeed in creating the assembly and having it moving under constraint with A2+ or assembly 3. I almost succeeded in getting it to work with assembly 4 and the animation of the master sketch but the assembly exploded during animation. Probably due to inconsistent signs in my master sketch. This could probably be solved with more time. In the end, I still don’ know which workbench I feel the more comfortable with! -_-

I know that the assembly I tried to model could be done easily by experts on any of the 3 assembly workbench. I did not spend a lot of time doing tutorials as from my experience, I thought it would be quite simple. A lot of « - »  points could probably be solved with research and tutorials. But as I said at the beginning, this is just thoughts from a regular CAD user with its habits that tries something new. I think that with a unique workbench taking good points of every workbench and solving issues that would remain, Freecad would be more easy to get in for beginners or professional users of commercial CAD softwares that want to use open source softwares for personal use.

Thank you for the amazing work you do! I cannot (yet) contribute with code but I hope this can be my small part to help developing Freecad.
User avatar
Zolko
Posts: 649
Joined: Mon Dec 17, 2018 10:02 am

Re: Assembly3, A2plus, Assembly4? Get united!

Post by Zolko » Sat Feb 22, 2020 7:37 pm

frabice wrote:
Sat Feb 22, 2020 12:39 pm
Hello,
Thank-you very much, Fabrice, I've been hoping for such a side-by-side comparison for a long time. I won't defend Assembly4 (yes, it can do sub-assemblies, it was actually designed for that) but I'd abuse your time a little more: could-you also stress-test the assemblies ? Imagine a scenario where you did the design of an assembly (or a sub-assembly) and you're handing it over to a colleague to open it, and to modify it and/or to assemble it in a greater assembly.

- What if the colleague doesn't have the corresponding WB ?
- What if he (or she !) modifies a part ?
- What if he moves the parts/assemblies to directories ?
- ...

You're analysis did already give very valuable information, thank-you for that.
try the Assembly4 workbench for FreCAD v0.19
install with Tools > Addon Manager > Assembly4 — tutorials here and here
catman
Posts: 156
Joined: Fri Jan 11, 2019 10:42 pm

Re: Assembly3, A2plus, Assembly4? Get united!

Post by catman » Sat Feb 22, 2020 9:58 pm

frabice wrote:
Sat Feb 22, 2020 12:39 pm
Hello,
I do not intend to get a response on each point from developers but just give my impressions. They can be used or not, it is up to everyone.
Thas really a very good side by side. I am very impressed about how many core key points you have in your list made in a very a short time. Many points are discussed somewhere in very long threads in this forum. I have not seen a place where they are summed up or maintained, so its very nice to get this snapshot view.
As you probably know the Assembly3 and 4 are not really finished and taking their first steps compared to the other software you are using, so your approach is a bit apples vs pears. Imho its worth pointing out what you do not have on your list: crashes, loss of work, erratic behaviour, missing documentation, etc - stuff you normally see when dealing with rather new features.

Maybe you would be able to do the same again in say, one year? Would be very interesting how things have changed by then :)

I think the three workbenches have so different approaches that its a huge advantage from user point of view to be able to highlight strong points in each one hoping to have them made available when using any of the other.
frabice
Posts: 28
Joined: Mon Jun 09, 2014 6:21 pm
Location: Clermont-Ferrand, France

Re: Assembly3, A2plus, Assembly4? Get united!

Post by frabice » Sun Feb 23, 2020 9:03 am

Zolko wrote:
Sat Feb 22, 2020 7:37 pm
Thank-you very much, Fabrice, I've been hoping for such a side-by-side comparison for a long time. I won't defend Assembly4 (yes, it can do sub-assemblies, it was actually designed for that)
Thank you for the answer, I will check that out.
Zolko wrote:
Sat Feb 22, 2020 7:37 pm
but I'd abuse your time a little more: could-you also stress-test the assemblies ? Imagine a scenario where you did the design of an assembly (or a sub-assembly) and you're handing it over to a colleague to open it, and to modify it and/or to assemble it in a greater assembly.

- What if the colleague doesn't have the corresponding WB ?
- What if he (or she !) modifies a part ?
- What if he moves the parts/assemblies to directories ?
- ...
These are good questions but I have to admit that I will not go that far for the moment. If I were working with someone, we would have the same tools (ie same FC version and same workbenches). After that, it would be careful collaboration on edits and file moves.
Concerning stress-testing the assemblies, I managed to break all of them by myself :D . That would probably be fixed if assemblies were done in a better way with more time on it. But at the moment, I do not find them very beginner proof :). Execption could be made for assembly 4 because of the very specific workflow of master sketch that let you few possible ways of "not doing the right thing".
catman wrote:
Sat Feb 22, 2020 9:58 pm
Thas really a very good side by side. I am very impressed about how many core key points you have in your list made in a very a short time. Many points are discussed somewhere in very long threads in this forum. I have not seen a place where they are summed up or maintained, so its very nice to get this snapshot view.
Thank you :).
catman wrote:
Sat Feb 22, 2020 9:58 pm
As you probably know the Assembly3 and 4 are not really finished and taking their first steps compared to the other software you are using, so your approach is a bit apples vs pears. Imho its worth pointing out what you do not have on your list: crashes, loss of work, erratic behaviour, missing documentation, etc - stuff you normally see when dealing with rather new features.
That is true!
catman wrote:
Sat Feb 22, 2020 9:58 pm
Maybe you would be able to do the same again in say, one year? Would be very interesting how things have changed by then :)
For sure. I would hope to have only one workbecnh to test next year. It would be simpler to point out things and concentrate on only one "product" ;).

Doing this job made me think about a way of doing assemblies that would be a mix between A2+/assembly 3 and assembly 4. Would it be possible to use transformations of local coordinates but thanks to geometries instead of placed LCS? That would avoid the use of a solver and make the thing more robust. For instance, if you put an axial coincidence between to cylinder, that it would make the two axis colinear but without involving any optimisation problem to solve. With this kind of approch, one could have the advantages of ease of use of geometric constraining and the rosbustness of not using a solver (or just a little to check if constraints are consistent). It is just an idea and I do not know if that is possible at all. Maybe it is already how WBs work at the moment.
Post Reply

Who is online

Users browsing this forum: No registered users and 2 guests