This is one of the very few times I've seen someone actually understand this topic. Most people talk about it as if it were a bug, rather than the simple fact that computers aren't able to read your mind.
great video, but the fact that problem can't be solved fully doesn't mean that we can't make it easier to work with. The improvement I am thinking of is to give user choice. I mean, when you do some change that break geometry, there would be selection menu on the side of the program to show user options to choose from. So default behavior could remain intact but application would predict most likely solutions to choose from.
Great breakdown! Very easy to follow. As a software engineer, this almost feels like when you change a software API interface and expect the consuming code to know where things moved - some changes just don't work that way. You can't automagically know how to update references.
It's quite similar I think. With sufficient abstraction I suspect it's exactly the same problem. Software development is a significant part of my day job. It's also why CORBA didn't solve any real world problems and faded away.
one way to improve this would be to add a popup option menu when performing an action which breaks geometry like this, for example "this action breaks the existing sketch , would you like to 1: copy original surface to attach sketch (default), 2: attach sketch to the nearest surface, 3: manually select a new surface to attach sketch, 4: manually fix sketch(do nothing)".
Great video! I've been using FreeCAD for at least 8 years, although just for periodic hobby projects of varying complexities. I learned a couple new techniques from your video. I wasn't familiar with ruled surfaces. I've probably used the split edge tool, but it's not something that I use often since I usually just know the geometry that I'm planning to draw and construct it line-by-line if it isn't a simple primitive like a square or circle. I've also never come across "inertia mode" for attaching a sketch to a face. I've always attached to the plane and then constrained relative to the x/y/z origin, but I can see how attaching with inertia mode would be a lot easier and more convenient in many cases -- especially when attaching to faces that aren't parallel with the XY, XZ or YZ planes. Otherwise, I fully agree with your commentary on the TNP, and I think the behavior exhibited by FreeCAD in this video is entirely appropriate. In the past, my greatest frustration with TNP issues in FreeCAD was when I had to rework a broken geometry but for some reason FreeCAD could not recompute my model when I tried to reattach sketches to different faces or otherwise rework the order and construction of my part designs without deleting some of the work and doing it from scratch in a different order. Hopefully FreeCAD's ability to reattach sketches to different faces without breaking has improved in 1.0.
CATIA V5 operates in a very similar manner. If I were building a model intended to handle a lot of parameteic and/or topological change I would be avoiding topological references as much as possible. Most stable - Reference geometry (explicit points,lines,planes), faces, edges, vertexes - least stable.
Very good explanation that the TNP is not really a problem of FreeCAD only, but basically affects every CAD programme. Without mentioning other programmes by name. I like it.
IMO the Topological Naming problem is still a problem. Ok, we all have our means to avoid it, but it costs time (is much faster to select external geometries, reference sketches to faces than having everything parametrized to tables and variables). As you said, it is impossible to the CAD figure out for itself our intentions, but it could have tools to help us. Something like a list of the referenced primitives that were renamed so we can check/fix, a option to replace external geometries in sketches rather than deleting and constraining everything again... things like that.
I agree. The TNP as it stands is more mitigated than it was in the past. However, the tools for resolving it when it occurs are not very straightforward. Also this example is a pretty extreme way to break a model (and no CAD package will be able to handle it perfectly), but there are much subtler operations that will break models in FreeCAD that are much less problematic in other packages I've used.
Excellent demonstration, unfortunately most people's attitude with modern complex software is not about: "How can I solve this problem with the free tools at hand?" but most like: "This is an inconvenience!, I refuse to put any effort to understand how to solve any problem or limitation in this free software because some other solution exists in a commercial package".
In this case, the commercial packages will have similar failures due to the nature of the problem. Some of them do not make recovery as graceful as FreeCAD does.
This problem is not really the same as the TNP. And yes any cad program would error out when the elements you are referencing are substantially changed or eliminated! When you saw the leg off a ladder, surprise, it falls over.
Prior to the mitigations, the cylinder would have ended up stuck to whatever random face got numbered as Face3 and no error flagged. Part of the mitigation is detecting that that is very unlikely to be helpful or anything like what the user wanted.
There was an effort by some to make the point TNP is endemic and can never be "solved", only mitigated to a tolerable level. Any time the term "solved" or "fixed" was used in print, one or more people would make a point of saying: "no, it is mitigated, not solved". I wish there had been better expectation control.
There was an attempt. I'm hoping that if I actually show a case that can't be fully mitigated and WHY, perhaps the appropriate expectation will get around.
- Nice example. Well explained :) - On a side note: though you were joking (haha), it is not the world's worst hammer - "The world's best toothpick makes the world's worst hammer." (Or something like that; and here, too, just joking :D ) - Carry on...
I rarely encountered it, but not never. But with the mitigation in place, I find that I can loosen up my workflow a bit and sometimes use attachments that are more appropriate without fear of something breaking as a result.
Ooh my cat can totally beat up your cad! Just watch! (I'm kind afraid of that psychotic flailing ball of fur and nails. It makes the most god awful sounds :( ). All kidding aside, thanks for explaining it :) .
This is a big deal, might have to look at freecad again, I am dying to get off of Fusion360, and I am willing to pay upwards of $75 a month USD to do so. Also better CAM software would be massive.
This is one of the very few times I've seen someone actually understand this topic. Most people talk about it as if it were a bug, rather than the simple fact that computers aren't able to read your mind.
That's because we don't have neuralink yet
great video, but the fact that problem can't be solved fully doesn't mean that we can't make it easier to work with. The improvement I am thinking of is to give user choice. I mean, when you do some change that break geometry, there would be selection menu on the side of the program to show user options to choose from. So default behavior could remain intact but application would predict most likely solutions to choose from.
That might make a good enhancement in future releases.
Great breakdown! Very easy to follow. As a software engineer, this almost feels like when you change a software API interface and expect the consuming code to know where things moved - some changes just don't work that way. You can't automagically know how to update references.
It's quite similar I think. With sufficient abstraction I suspect it's exactly the same problem. Software development is a significant part of my day job. It's also why CORBA didn't solve any real world problems and faded away.
one way to improve this would be to add a popup option menu when performing an action which breaks geometry like this, for example "this action breaks the existing sketch , would you like to 1: copy original surface to attach sketch (default), 2: attach sketch to the nearest surface, 3: manually select a new surface to attach sketch, 4: manually fix sketch(do nothing)".
That could be a nice enhancement now that there is a good base to work from.
Great video! I've been using FreeCAD for at least 8 years, although just for periodic hobby projects of varying complexities.
I learned a couple new techniques from your video. I wasn't familiar with ruled surfaces. I've probably used the split edge tool, but it's not something that I use often since I usually just know the geometry that I'm planning to draw and construct it line-by-line if it isn't a simple primitive like a square or circle. I've also never come across "inertia mode" for attaching a sketch to a face. I've always attached to the plane and then constrained relative to the x/y/z origin, but I can see how attaching with inertia mode would be a lot easier and more convenient in many cases -- especially when attaching to faces that aren't parallel with the XY, XZ or YZ planes.
Otherwise, I fully agree with your commentary on the TNP, and I think the behavior exhibited by FreeCAD in this video is entirely appropriate. In the past, my greatest frustration with TNP issues in FreeCAD was when I had to rework a broken geometry but for some reason FreeCAD could not recompute my model when I tried to reattach sketches to different faces or otherwise rework the order and construction of my part designs without deleting some of the work and doing it from scratch in a different order. Hopefully FreeCAD's ability to reattach sketches to different faces without breaking has improved in 1.0.
CATIA V5 operates in a very similar manner. If I were building a model intended to handle a lot of parameteic and/or topological change I would be avoiding topological references as much as possible. Most stable - Reference geometry (explicit points,lines,planes), faces, edges, vertexes - least stable.
Very good explanation that the TNP is not really a problem of FreeCAD only, but basically affects every CAD programme. Without mentioning other programmes by name. I like it.
Thanks for this video. I was so confused the last days when I kept reading that FreeCAD had “solved” the TNP.
IMO the Topological Naming problem is still a problem. Ok, we all have our means to avoid it, but it costs time (is much faster to select external geometries, reference sketches to faces than having everything parametrized to tables and variables). As you said, it is impossible to the CAD figure out for itself our intentions, but it could have tools to help us. Something like a list of the referenced primitives that were renamed so we can check/fix, a option to replace external geometries in sketches rather than deleting and constraining everything again... things like that.
I agree. The TNP as it stands is more mitigated than it was in the past. However, the tools for resolving it when it occurs are not very straightforward.
Also this example is a pretty extreme way to break a model (and no CAD package will be able to handle it perfectly), but there are much subtler operations that will break models in FreeCAD that are much less problematic in other packages I've used.
Excellent demonstration, unfortunately most people's attitude with modern complex software is not about: "How can I solve this problem with the free tools at hand?" but most like: "This is an inconvenience!, I refuse to put any effort to understand how to solve any problem or limitation in this free software because some other solution exists in a commercial package".
In this case, the commercial packages will have similar failures due to the nature of the problem. Some of them do not make recovery as graceful as FreeCAD does.
This problem is not really the same as the TNP. And yes any cad program would error out when the elements you are referencing are substantially changed or eliminated! When you saw the leg off a ladder, surprise, it falls over.
Prior to the mitigations, the cylinder would have ended up stuck to whatever random face got numbered as Face3 and no error flagged. Part of the mitigation is detecting that that is very unlikely to be helpful or anything like what the user wanted.
Nice. :)
Very well done, so I subscribed. Thanks.
thanks for explaining that better.
Nice explanation and solution.
There was an effort by some to make the point TNP is endemic and can never be "solved", only mitigated to a tolerable level. Any time the term "solved" or "fixed" was used in print, one or more people would make a point of saying: "no, it is mitigated, not solved". I wish there had been better expectation control.
There was an attempt. I'm hoping that if I actually show a case that can't be fully mitigated and WHY, perhaps the appropriate expectation will get around.
The title is misleading, but it is true. FreeCAD had MITIGATED the issue to make it AS GOOD as other car platforms (most of the time)
Excellent explanation!!!
Topological naming problem can not be fully solved. Only mitigated to a certain degree.
- Nice example. Well explained :)
- On a side note: though you were joking (haha), it is not the world's worst hammer - "The world's best toothpick makes the world's worst hammer." (Or something like that; and here, too, just joking :D )
- Carry on...
This is the behaviour I expect. In NX it is solved the same.
Exactly!!! This is NOT an issue unique to FreeCAD ... The FreeCAD authors hav never said the problem was solved only mitigated
Exactly. This is the best that can be expected. It's great that FreeCAD has advanced to be on-par in this area.
I've never had TNP in Freecad with my workflow
I rarely encountered it, but not never. But with the mitigation in place, I find that I can loosen up my workflow a bit and sometimes use attachments that are more appropriate without fear of something breaking as a result.
Thank you! 😎👍
You're welcome!
2:00 worlds worst hammer 😂
best.... subcribe
Ooh my cat can totally beat up your cad! Just watch! (I'm kind afraid of that psychotic flailing ball of fur and nails. It makes the most god awful sounds :( ). All kidding aside, thanks for explaining it :) .
This is a big deal, might have to look at freecad again, I am dying to get off of Fusion360, and I am willing to pay upwards of $75 a month USD to do so. Also better CAM software would be massive.
wow fantastic😂