All datums changed position/angle/location

[Pauvres_honteux]

Topological naming limitation. You attach something to a face/edge/vertex etc. of an object, but when you edit that object what used to be Face 2 may well now be face 3 or anything else.

Don't attach datum planes etc. to generated geometry, doing so is quick and usually pointless

Shouldn't the

Bance has created his widely recommended Basic Attachment Tutorial.

tutorial then be, at least, rewritten?

[Bance]

tutorial then be, at least, rewritten?

Can you describe where in the tutorial it is recommended?

And this should really be in the Tutorial discussion thread-

https://forum.freecadweb.org/viewtopic.php?f=36&t=53095

[Pauvres_honteux]

Can you describe where in the tutorial it is recommended?
https://forum.freecadweb.org/viewtopic.php?f=36&t=53095

Sure, search for "Normal to edge Isosceles" and "Normal to edge Attachment" and "it should be constrained to the origin at the vertex adjacent the hypotenuse".
I think you'll find more places where edges and vertexes are effectively turned into parents for something.

[jmaustpc]

Can you describe where in the tutorial it is recommended?
https://forum.freecadweb.org/viewtopic.php?f=36&t=53095

Sure, search for "Normal to edge Isosceles" and "Normal to edge Attachment" and "it should be constrained to the origin at the vertex adjacent the hypotenuse".
I think you'll find more places where edges and vertexes are effectively turned into parents for something.

Hi guys
I had a very quick look through that tutorial, you may have found an error or not, I did not look closely, but I just wanted to say that the tutorial clearly states that it is only a simple demonstration of how attachment works. It says that it is not intended to be the best way to model the object and clearly warns about the lack of robustness in this specific example. So at least in general, it seems okay to me. I will let Bance deal with the detail you refer to but feel free to ping/quote me if either of you do want my detailed opinion/advice.

Jim

[Bance]

Ping

Jim care to split this thread, move the last three or four posts to the tutorial discussion thread.

So far as I can see Jim is talking about generated geometry, none of the quoted points are generated geometry.

They are all explicit, forming part of a sketch. Your third example is one of constraining a sketch to the origin (if that is wrong then, exactly how would you fully constrain the sketch?)

As noted above, it is clearly stated that using the methods described in the tutorial does not guarantee the absence of TNP, merely that it follows the guidance given in the Wiki for making stable models.YMMV.

[jmaustpc]

Jim care to split this thread, move the last three or four posts to the tutorial discussion thread.

No worries.

[Pauvres_honteux]

A constructive and foreward/upward striving suggestion could be to outline/teach newcommers to not use edges or vertexes originating from surfaces or solids as parents for anything, if possible. And set it as the headline for the tutorial.
Add this sentence directly under the headline: As of todays date it is not possible to get around the topo-naming problem, one can only make the model a little less affected by it.
As it stands today, and for this tutorial, we must use what FreeCAD denominates edges and vertexes, but not edges and vertexes from surfaces or solids.

And amend that topo-naming solutions, what ever they are and with high probability, will still not make models stable (even in the future) when your model increases in complexity and/or more changes is made over time.

Also mention that stable models does not exist, one can only make less crash prone models. And stress that as of todays date the "crashability" is in the lower part of the "stable scale".
With realthunders take on toponaming in the comming months we will able to make models which can be attributed to the upper part of the stable scale, but they will never ever be fully stable.

To advance this further I suggest we make a model (perhaps the same shape as in the tuturial?) which gets us as high up as one can get on the "stable scale", as of today. The modelling technique will revolve around points(not line ends) in the sketcher made selectable from outside the sketcher.

[Bance]

Thank you for your input, the tutorial is about demonstrating attachment not making stable models.

I have tried to adhere to the recommended workflow. If you find there is anything wrong in the tutorial, please advise and I will amend.

I rather think that your opinions would be better placed in the Feature_editing page or perhaps the Topological_naming_problem page.

Feel free to update the feature editing page (stable models section) with your recommendations, or even create a new page.

I will take these points into consideration, should I decide to add to either of those pages (I have been thinking about it.)

[Pauvres_honteux]

I do not think you can decouple attaching stuff from the topo-naming problem.

To not give new users false hope I encourage you to enlighten them, and even stress, the topo-naming problem in the attachment tutorial, since attaching IS the very start and root cause of topo-naming problem, seen from users perspective.

The fact that the programming aspect of topo-naming problem really stems from OCC-code being utterly poorly documented might be mentioned as a side note.

Then give them a light in the darkness by mentioning some salvation in the form of His Highness Lord Thunder ongoing implementation of his take on topo-naming.