Constrain solving - as complicated as that

[Cris-Ideas]

Hello All,

I have started this thread to bring up and hopefully explain problems related to constrain solver and gain knowledge on how exactly constrain solver handles the assembly.

This is because I am experiencing many problems related to assembly solving that should not be expected.

Originally I was discussing this issues in the thread related to problems with flexible assemblies but I decided to split this topics and this thread is intended to focus on constrain solver it self.

I would like to encourage especially people from Autodesk who have the knowledge about constrain solver it self to participate and help us understand better how this works to make our lives easier.

I will be posting videos and data sets, so anyone could try the same what I do and see if is getting the same outcome.

I also would like to ask you to do the same, and if possible use videos with comment rather than long posts, as this is more easy to follow and understand the intent.

I will be also giving each example a name, so it was easy to track replays. So when posting please make it obvious what you are referring to. 

If you are interested in problems related to flexible assemblies please visit this thread Flexibility not working properly in inventor - BUG that has been there for ever

Cris.

[johnsonshiue]

Hi Guys,

Let's see what solver team says. Like I said, the inconsistent behavior is indeed wrong. It is either a bug or a limitation. It should be consistent (both fail or both work). The redundancy analysis on or off should not change the result.

Regarding having an Insert constraint between the two washers, indeed, it does help this case, assuming that all the Axial-Axial Mates are Undirected (Opposed or Aligned would not allow flipping here). Maybe the solver detects that flipping Mate:10 is an acceptable solution.

Please do not quote me. But, I have a feeling that this particular flipping behavior should be disallowed (or fail) in the first place. So, the bug is really about allowing it during quick solve (redundancy analysis off).

Many thanks!

[Curtis_Waguespack]

---

@johnsonshiue wrote:

Please do not quote me. ...

---

Too late 

Maybe I should have waited to order 12 of these?

I'm joking of course.... everyone have a great weekend!

[johnsonshiue]

Hi Curtis,

I got to get one of these. This is too cool! I have a dream, that every word I said would be printed on every shirt and embedded in every mind. I got to print it out and hang it on my cube.

Many thanks! Have a nice weekend! It has been a very long week for many of you!

Thanks again!

[Cris-Ideas]

Hi All,

How about #166 bug and this constrain becoming redundant after pattern is created.

Any comments, thoughts, explanation maybe?

Cris.

[Cris-Ideas]

I see that example with constrain becoming redundant after pattern is created did not get much focus.

So here is #169

another example of obvious bug in the software.

data set available for download:https://autode.sk/2Mzv6M0

Cris.

p.s. I already know where this bug is but finding it will be most surprising for you. I expect at least few videos "proving" this works. I can than tell you which option to change in your inventor so it also give the error. According to how Autodesk's solver works (or does not in fact) this assembly will bring the error. If it does not, it is only by lucky coincidence.

[johnsonshiue]

Hi Cris,

Regarding #172, I have worked with the solver team to understand the behavior better. It is indeed a solver bug, reported as INVGEN-19140. When "Enable redundant relationship analysis" option is on, the behavior should be consistent with the option off.

Many thanks!

[Cris-Ideas]

Hi Johnson,

Thank you. But you have stated the obvious, that is:

       That the behaviour of the software should be consistent.

But could you please give us some answer to a little less obvious questions:

1) Witch of the current behaviours is according to what is expected from the software.  (suspect what the answer will be, but it is better to have this stated)

2) When should we expect that inventor will present this consistent behaviour?

3) Will bug shown in #169 be fixed in the same time?

Cris.

[johnsonshiue]

Hi Cris,

I believe the desirable behavior should be the solved behavior, not the failed behavior. The failed behavior is a bug. Regarding when the fix will be available, I am sorry I cannot provide an estimate. I can only hope it will be soon, since the case is straightforward and it should not be too hard to debug. But, you never know. If a change introduces other undesirable behaviors, we will need to find a different solution, which will take longer.

Regarding #169, I don't think it is completely solver related. It has something to do with how Inventor groups component pattern. Component pattern is a bit different. The pattern as a whole can move together. However, the patterned occurrence cannot drive pattern source. I suspect this case might be due to the fact that the constraint is trying to push the pattern occurrence, which could drive the pattern source leading to failed constraint. The behavior is confusing regardless. I will work with the project team to understand it better.

Many thanks!

[Cris-Ideas]

@johnsonshiue

@#169

Driving pattern source by constrains applied to other pattern occurrences is perfectly doable.

Also if constrains are allowed for pattern occurrences this MUST be the case if whole assembly is to be solved. There is no other option to CALCULATE solution other than allow occurrence to drag master element of the pattern and whole pattern with them. (currently in inventor this is not the case)

In this particular case constrain that I try to apply is perfectly fine, as pattern as whole, and pattern's master occurrence have free DOF that allows for this constrain to be followed.

This particular example is very good because it will allow you to find very interesting bug that is causing solutions to be totally unstable, as it can be demonstrated that in different circumstances and different clicking workflow this particular example does not bring up the error message and "seems to be calculated" but in fact it is only "solved" by accidental chain of actions. So it is as solver was walking around blind and just happen to hit the solution.

Have you also looked @ #166?

This one is also very interesting and not less surprising. How a constrain can become redundant when only change in the assembly is so that component was patterned? 

There IS NO theoretical justification for such a result. It is TOTALLY WRONG.

@johnsonshiue

As you already know I had done my homework.

I had worked out the theory on how to solve such geometrical structures as assemblies made of constrained elements. It comes out straight from the theory so there are TWO, AND ONLY TWO possible solutions for a assembly solve process.

• Either assembly is solved          OR
• There is no solution because some constrains are in conflict.

It does not matter how many free DOFs there are, or how heavily over-constrained assembly is.

So

Basing on the theory there is no justification for unstable solver behaviour.

I must herby admit that my speculations I placed in post #33 are almost totally wrong.

It turned out that solving of the assembly is far more easy than I had ever expected. But that is often the case when one finally realises how something is done.  There are in fact very few equations to be solved ad further more all are linear.

It is amazing how math allows to solve so "complicated" problems in fact basing on very simple mechanisms.

@johnsonshiue

Hope you will explain #166 and #169.

Cris.

[Cris-Ideas]

---

@johnsonshiue wrote:..
...... If a change introduces other undesirable behaviors, we will need to find a different solution, which will take longer.....

---

@johnsonshiue

This question does not want to go away:

How solving something that is obviously wrong, and what concerns very basic functionality, and effectively one of the foundations of Inventor can potentially lead to "other undesired behaviours"?

I consider this a rhetorical question, but it seems it needs to be stated.

I would not like to come to a conclusion that this undesired behaviour is users being content and finally having software they do not have to struggle with every day and do not have to effectlessly wait for a next update or relies, hoping problems that cost them time will finally be fixed, after so many years of paying for the software that should do that is was designed to do, but it still it does not do that.

This single constrain solver buggy behaviour had made me alone lose few months of my life for struggling with assemblies foaling apart, POS reps I was not able to get the the top level assembly, designs being destroyed just before I had them finished, documenting all of that and sending support cases, struggling with Customer Support trying to explaining them that "suppressing and un-suppressing constrains that are perfectly fine is not the solution in this case", on building numerous assemblies for the same element only to finish the job not being able to use simple flexibility functionality.

This just came two days ago, AGAIN:

..................................................... "

............................................"

@johnsonshiue

Can you imagine how frustrating it is when only what I get is:

---

@johnsonshiue wrote:..I am sorry I cannot provide an estimate. I can only hope it will be soon, ..

---

This is just not good enough!

I have been hoping it will be soon, for 8 last years. And you know what?

I did not work!

Hoping does not get things done.  - Getting things done gets things done!

If you cannot promise us anything give us e-mail to someone who can.

Cris.

[johnsonshiue]

Hi Cris,

I would like to stick to technical facts. The information I provided so far is the truth I know. I don't think anybody can give you promise when any given issue can be fixed. It does not matter who you talk to. Please feel free contact any Autodesk employee you know. The answer will be the same. I am not liable to provide additional email addresses for you.

Regarding #166 and #169, I do not have the files on my machine and I don't recall you sent them to me. You will need to show me which cases you reported against. I can contact our support team to get the files.

Many thanks!

[Cris-Ideas]

Hi Johnson,

# 166 and # 169 are described earlier in this post.

videos and links are provided.

But I also give them here.

#166

#169

And as for when things are done.

I am sure there is a plan what new functionalities are to be introduced in the next relies. And surely there is someone who decides, assigns resources and expects results.

This is most likely Inventor Project Management. But I was not able to track this e-mails yet 😉

Cris.

[Frederick_Law]

Don't shoot the messenger.

[johnsonshiue]

Hi Cris,

Many thanks for sharing the links! I will take a look and get back to you.

Regarding how Inventor team is run or how the company is run, I personally do not believe it is something to be discussed on a public forum. Autodesk employees are encouraged to interact with users in various channels.  The objective is to understand users' requirements better and deliver better experience. You, as a user or a customer can criticize the delivered solutions. But, I personally think it is very inappropriate to comment on the personnel, processes, and internal matters, which are all irrelevant to the issue and to the user base. It is as if an outsider tries to drive how the product team and the company works. I don't believe you would like somebody else tells you how to run your business.

You are entitled to your opinion and you are free to express however you like. It is your choice. You are protected by freedom of speech. I get that. I just want to remind you that there is a level of trust and respect within the forum. I try my best providing technically accurate information but I reserve the right to provide any information. I don't have to fulfill every request you ask. You are not the only user we serve and interact.

Many thanks!

[johnsonshiue]

Hi Cris,

#166 looks like a bug to me. I cannot explain why the Mate constraint becomes redundant. It should not be. Did you report it as a case? If yes, what was the case ID?

For #169, I cannot reproduce it on 2017.4.5. It may have been fixed. It looks like you are on 2017.3. Please install 2017.4 followed by 2017.4.5 update.

Many thanks!

[Cris-Ideas]

@johnsonshiue

Hi Johnson,

I have reported both of them

#169 is reproducible on your machine. I can explain you how but I must know you application options  and screen cast from your attempt.

Could you provide those?

Cris.

[johnsonshiue]

Hi Cris,

I think I see it now. The preview button needs to be unchecked in order to see the bug. It is wrong. Yes, this is a bug. Please let me know what case IDs you have for #166 and #169. I will work with Support team to see if they have been reported as defects.

Many thanks!

[Cris-Ideas]

Hi,

#169

I see you did  your homework on this case.

As you can see it is only solved if first components are positioned and actual solve is done after. So this is not the solver that is positioning this elements.

support cases ID are:

CaseNo:14463540.

Case ID: 12666540

Cris

[johnsonshiue]

Hi Cris,

#169 was reported as INVGEN-19011. It is under investigation. #166 was reported as INVGEN-5051, which has been fixed on an internal build targeting future release.

Many thanks!

[Cris-Ideas]

@johnsonshiue

Hi Johnson,

It was not my intention to comment on how Autodesk is run.

But the thing is that as a user and customer I am struggling with software that does not do what it is said it should.

I am constantly logging issues with customer support to ally your team to know how to reproduce them.

I am also constantly posting on forum as a second channel of communication that has this benefit that is public and I can get help and support from other users. Also I will have access to it when not on subscription.

All what I can expect is information "thank you for reporting that, it has been logged, ... we can not promise any date..."

As previously said. In my opinion this is just not enough. As a customer I expect much more solid answer in regard to reported bugs.

This bugs cost me time and money, and additional time I spend documenting them. I would expect that they are assessed and planed to be fixed in set time, or decided they will not be fixed.

If you can not promise anything, as a customer I would like to get contact to someone responsible for this matters.

I would like to place my concerns and feedback directly where it may have some effect for me.

Technical issues I place here, directly to you and through customer support.

But in terms of deadlines no one can promise anything. But is is obvious there is someone in the organisation who can. So I see no reason why he or she would not like to have a conversation with me.

I think by amount of feedback I  have provided I have proven it is not my goal only to criticize but to act in a way so this was productive and allowed for improvement.

But I really expect more than "I cannot promise".

Cris.