saving doesn't `stick'

Generally when a general fault crops up your model is toast no matter what you do. No pun intended.

Untrim the face that is causing the problems as well as adjacent faces. That might reveal what is going on.

If you can save out enough of the imported geometry surrounding the bad area to a temp file maybe you can fix it and reimport and replace the faces.

What is the source for the bad imported geometry.

What's wierd is that a ctrl q seems to fix it-temporarily. Also after doing this and opening up the derived parts, there are no problems until I save and close and the re-start the next session. Also, I was checking it all the way through as I was creating the geometry and no problems cropped up until I saved and closed. Then I saw problems when I started a new session that I had never seen before.

The original files were Pro-E native which SW imported with no problems. Looks like whatever I did to create the parting line surfaces is the source of the problem. Not sure I can blame Pro-E for this one.

jk

Given that SW can create and export geometry that even it can't import all this doesn't surprise me.

It seems like what is in memory isn't getting written to disk or isn't getting read back in from disk.

Still I would have a good look at the Pro/E geometry and SW interpretation of it. Untrim may be very helpful.

Also, remember that SW doesn't always complain when it should. Over the years there have been several times when SW didn't report errors when they were in fact there. Then in a next release the error checking was put in place and people complained that the new release caused errors when in fact the old release didn't report them. 2001+ to 2003 comes to mind.

Are you using verification on rebuild all the time? Also are you using the feature option in the Check tool?

yes

Also are you using

Yes. Also, I was checking every step of the way. All was well until I saved it and closed. The general faults first appeared after reopening the file. After the first time, I went back and rolled back and checked every feature, fixed a few and saved again. A tools , check reported no invalid faces or edges. Then saved this apparently good file and exited SW. Start a new session and the general faults are back again. The control q works every time and I am nearly done with the mold design. However, it takes me about

20 minutes every session to clean up the split and then rebuild the child assembly with all of the derived parts.

I made extensive use of the unbtrim tool when building my parting line, but the way. I really like that feature.

jk

"John Kreutzberger" a écrit dans le message de news: snipped-for-privacy@corp.supernews.com...

Does saving to parasolid and importing back an option?

"John Kreutzberger" a écrit dans le message de news: snipped-for-privacy@corp.supernews.com...

Is saving to parasolid and importing back an option?

That is exactly what I am doing today. I lose parametrics, of course, but at least I can finish the job this way.

Making changes requiring rebuilds of multiple parts was getting to be out of the question anyways.

Interesting thing about round-tripping the parasolids file was opening up every part and running diagnosis of the imported parts. Many of them had bad faces, but they were all fixable. Seems like SW was creating bad geometry and not catching it-even with `verification-on'.

In response to Stefan below-I do think this is a bug and thanks for the confirmation. I am sending the whole lot to my VAR today.

jk

I've seen this about a handful of times since 2001 and I'm guessing it's related to tolerance adjustment in the parasolid kernel? One way I could resolve the problems was to add/remove possible problem areas before any features to the imported body or exporting in IGES or STEP and again try to find any possible edge or surface which may cause a problem later. For parts which were created natively, I would cut/split or trim away areas which were suspect. Doing a parasolid out would sometimes help but not always, the BREP or edge tolerance seem to be the problem?

BTW, one simple test (as you probably are aware) or sign on problem bodies is using the scale feature, I found that doing a scale of 0.0 would fail on problem bodies or native created parts. Or, (I'm sure some of you may have seen this before?) if you bring in part(s) which are significantly larger or smaller than the assembly or part referencing a imported body, the body or references to that body may sometimes rebuild!?

So, I think it has to do with parasolid or how SW chooses to use the body/edge tolerancing?

Anyhow, even though I could workaround some faults, some files though,.. after doing all the above, and doing ctrl-q's,.. would still on a later time during the edit cycle or sp or new release.

..

Interesting, Paul. I also suspect the parasolids kernal. The parasolids roundtrip didn't work on the 2 most complex parts, but step did.

jk

Did you also try VDAFS. That format has more stringent checking than STEP.