As severe as a problem can be. Total file deletions in entire working folder

I have left the dll’s operational so SketchUp is still able to access them and do its thing. What is disabled is the the report sender. That doesn’t kick in until a crash happens. I imagine if you removed the dll’s there would be a problem but I haven’t.

Until it is established what caused the file deletions I am not interested in putting it back in operation. Considering that BugSplat is the only part still running after a crash it can logically be the only part attempting to write things to the file system in random places that do not make sense.

I have been in the computer business for just over 50 years and for about 37 of those years it was not a hobby. I worked for a multinational. I am sorry if I seem unwilling to take advice but that is a bad habit I developed dealing with some customer’s IT departments. I am considering what has been suggested and will be trying working without the plugins (except one). The problem with bugsplat is not “negotiable” as I do not have another machine powerful enough to act as a test box.

BTW, there is one small clue about the file mapping attempt. It brought up the standard file mapping window but it had already inserted a folder name to use for the mapped folder.

That is “g but”. I have no such folder on my system (created by me). That is what makes me think that it is something somehow left in the BugSplat build or maybe SketchUp or even a plugin. “g but” sounds like a shortened form of “g button” to me. That just might mean something to someone.

How you tried removing the problem plugins files and retesting - as already explained to you by several of us ?

Disabling ‘some’ plugins - e.g. everything except ‘move 2’, will at least show you after some testing if ‘move 2’ is causing issues.

Do you have the Ruby Console open when testing - if there are error-messages they might be very informative…

Assuming ‘move 2’ is NOT the culprit… re-enable some of the usual suspects… re-test…
That way we might know what is happening.

To talk around the issue without doing what has been advised is not getting you anywhere nearer the solution.
You ‘know’ it’s BugSplat, so don’t look at other possibilities.
I not saying that it’s not BS, BUT you haven’t tested the other possibilities yet either…
There’s lots to do.
It’s up to you…

I will be working with it this evening (PST). BTW, as a comment there is a method in the BsSndRpt64.exe called “FlushFileBuffers”…

[quote]There’s lots to do.
It’s up to you…[/quote]

The fact that I am having such a problem puts it far beyond just me.

I am working on this. I don’t want to stop the crashes at this point. What I want to do is make it crash repeatedly and predictably using as few entities as possible and then try to create the smallest possible sample model that exhibits the behaviour. That is the best way to assist the programming team if it is an inbuilt SU problem and will also make it a lot easier to pin down other suspects. I have noticed there are some visible issues that make me sure it is going to crash soon, and it does. Those visible issues are very difficult to explain so I may just make a sequence of images showing them.

One thing I have been seeing is that SU is complaining that I cannot name two components the same even though it also says that there is only one in the model. Somehow connected to this is the appearance of what seems to be the same component a distance away but it has no visible edges no matter how you try to display them. Nor can it be selected with the cursor directly but it can be selected by outlining it. All the while SU maintains there is only one such component. When this happens a crash is imminent if that component is handled in any way including layer switching.

I removed all plugins except for Mover 2. I loaded a copy of the model that I was able to recover from the drive that I specifically labelled as one that was crashing. In fact, it was nearly all set up to do a move of a part using the standard mover tool so it was ready to test where it had crashed before.

It crashes every time with and without the mover 2 plugin installed. It isn’t a plugin problem.

Now the problem I have is that it must be done in the exact sequence with no changes made. I have tried just saving the objects being worked on, reloading just them, and that does not crash. An interesting thing about how this is happening is related to layers.

I am moving an object that is a fully unique copy of another. It consists of just two parallel planes near each other, a wall. The reason is to replace it. The one being replaced has serious planar breakup most likely caused by the transfer from ver 8 to 2015. I am quite sure there is a difference in the internal math precision causing 2015 to detect non planarity with smaller mathematical differences than in ver 8. This is highly amplified by the very large scale I am working at. The maximum dimension of single planes in this model is about 900 metres. Over 500 is common.

The replacement part has been remade to remove the myriad breakup hidden lines across the planes. It is an identical copy in all respects but it is not a direct copy of the original. The original is in a specific layer, the replacement is in Layer zero. The object to be replaced is the only item in its layer.

In sequence, the replacement is to one side ready to be moved along an axis to the location of the original part. A guide line is drawn to help with the move. The original is deleted. The replacement is moved by hand using the standard mover tool to snap into location at the exact same point as the former original.

Once the replacement is snapped in place the moment the mouse button is released 2015 crashes instantly.

Of interest, In this testing one time I didn’t quite get the replacement in the exact same place as the orignal. SU did not crash. I then did a few simple modifications to the replacement part and a message popped up saying that what I was doing had caused a change to a hidden part. The problem with that is that there was not supposed be a hidden part. It was deleted before the move was made. It was the only part in that layer.

There is an issue with the layers system. It is not keeping track of what is in a layer correctly. Even so, it shouldn’t crash if two parts are melded, even if it is hidden. Also, both parts had different component IDs.

The deletion of files issue may well be unrelated to this. I suspect it is a BugSplat problem.

At this point there isn’t much else I can do. I have a huge number of parts so not using layers isn’t a viable option. The ball is in Trimble’s court.

Where is your Model’s origin relative the the ‘parts’ - [0,0,0] - axes-crossing point.
You said something earlier about working ‘to scale’…
If you have parts made at one size but inserted at another scaled size then that might cause issues.
I have seen this only once [earlier SUp version], where a CAD author made a DWG in m, but had imported a door block that was drawn in mm as if it were made in m, so finding it 1000x too big they had then scaled that inserted block in the DWG [rather than scale the block’s contents], and then copied its scaled instance around the DWG. That DWG was imported into a SKP. This hyper-sized door component-definition made from the CAD-block was ~1km wide and although its instances were only ~1m wide, any work on the model involving those caused random splats. Simply scaling the component’s contents to the correct size, and replacing the instances solved the problem.
SketchUp’s lower tolerance is 1/1000" - any points closer than that are taken as coincident, so tiny edges and facets are not created - they can be made in a temporarily scaled up component and then scaled down again: tiny geometry can exist - it just can’t be created. SketchUp’s upper tolerance is not documented, but certainly it might have issues will ‘big-numbers’ in points and also have issues deciding on coincidence etc…
SketchUp was not set up to model the parts of swiss-watches, nor entire cities - it is comfortable with parts of buildings, buildings and city-blocks… although there are users succeeding with models of similarly sizes to yours, and with much bigger Mb’s…

Have you used Model Info Statistics to ‘Purge Unused’ from the Model ?
Have you uses the nearby ‘Fix Problems’ option ?
Is you raw-geometry on Layer0, with other Layers assigned to ‘containers’ - groups/component-instances ?

The model was imported exactly as I have been working on it for the last year. No changes have been made to scale. Origin is at the approximate centre of the entire collection. This is a model full of models. Many of the models are built on a “workbench” near the origin and them moved to the final location. I am sure that down scaling will eliminate an further issues with planar breakup but that will be the last step.

I am very familiar with the limits of accuracy in the previous versions of SketchUp but I haven’t yet done any experimentation on that directly in 2015. The entire idea with this level of upscaling is to make it possible to work at any level of detail that I wish. The scale is currently 1 inch scaled to a metre, 39.3701 to 1. Using that scale makes it easy to work in inches since both are handled by SketchUp. I am entirely comfortable with either system since I learned both as I grew up.

I purge the model regularly and check the validity frequently. All parts are on layer zero with only the final major collections on the management layers. These are all things that I learned long ago as I have been using SketchUp since version 4 and have made hundreds of models. Some are on the warehouse. Many are not because of the very small size limit that was in place for so long. I also use SU to design actual objects that I then make using my CNC milling machine, including laser cutting and engraving.

My file sizes are far lower than most because of they way I take advantage of components. I make components anywhere that symmetry exists. Something that is a square with four curved corners is then made from four components, one for each quarter of the window, for instance.

I haven’t yet decided how I want to post this model. As one giant collection of hundreds of models or as the primary model alone with each smaller one separately. The models range from a urinal to a pots and pans hanging rack. The main container is a small ship, a research vessel, all my own design. It is all designed using scientific reality as a guide. The design of the ship includes proper allowances for draft and displacement. The props are the correct size and pitch for the engines it would use. Every single thing could actually be built using my designs. I have worked in the machining business on the side all my life. My experience in computers is largely related to computer control systems for machinery which is why I know machine language well. Computer graphics is something I started with in about 1979 when I designed and built a high res video system to run on a Commodore PET.

https://3dwarehouse.sketchup.com/user.html?id=0031359538483141435820637

Just a couple of things to note from a very quick look at a couple of your models.
Keep an eye on your face orientation, perhaps they were older ones but it was quite random in the two I looked at.
Your Parabola is a great example of the incorrect use of layers and how it can cause problems.
You have the edges of the segments on a layer other than Layer0 and not even on the same layer as the faces. This could be causing your issues with the error about changing hidden parts.
If you try to move faces from one layer to another when their edges are on a different layer it will break things.

The array point of your parabola is an interesting collection of overlapping random points.

Go to Window > Model Info > Unit and check that the precision is still set as desired.

Custom templates from SU8 are at C:\Program Files\SketchUp\Google SketchUp 8\Resources\en-US\Templates. If you have any you want to transfer, open them up and re-save as SU15 through File > Save as Template. You can have more than one model style - one for fast modeling and another for presentation, among other needs. Save the different Styles to different Scenes and switch back and forth.

Just because your graphics card is decent, still doesn’t mean its latest driver is OK. Disable hardware acceleration through Window > Preferences > OpenGL to check. Though you may need to test in a smaller model.

Running SU in clean boot can help to eliminate if there are conflicting issues with your system while still using the NVIDIA driver.

The misuse of Layers as Box noted could have lead to some ov the visual issues you reported a few posts ago. In general, raw geometry on Layer0 and groups/components on different Layers.

Unless you are a Pro user, this collection of problems only you have reported isn’t Trimble’s issue. Besides, they look at bugsplats.

The precision is usually set to maximum and it doesn’t change.

Some of the items I have posted are rather old. Some go back a number of versions. In this case there is no issue with having items mixed up in the layers as I have been very careful to avoid that.

I create my own templates. They do what I want and need.

The fact that this is SketchUp Make in no way relieves Trimble of responsibility. There is no difference between the paid product and the free product. If the free product has problems that may result in losing all of your work how many people are going to be interested in actually paying for it??? I have submitted quite a few bugsplats already and it will not be re-enabled until I hear about some serious investigation into how all my working files were deleted. That is not something I have seen any reputable software do in the past. It isn’t a good way to gain new customers and I spend a lot of time on another forum where a large number of people listen to what I have to say about software suitable for new CAD users. Many of them are people who need it to make a living. At this point I cannot recommend using SketchUp 2015.

You have already had a ton of suggestions, and have turned them all down. You seem to have a clear conception of the suggestion that you want to hear, so what is it?

Anssi

I have not turned them all down, I have done exactly as suggested. Incidentally, there are no layer issues with the parabola. It displays just fine in 2015. It has just layer zero and one named default. Layer zero contains the framework and default contains the subdivisions. All 180 components are exactly the same. It is an extremely simple model.

I will probably have to cut off the conversation. I was just told they have room for me at the hospital so I will be spending a week in there.

Just trying to clear up this layering.
When people talk about raw geometry needing to be on layer0 and only groups and components on other layers they mean the actual edges and faces within a group or component.
You have the component on Layer0 and the faces within that component on layer0, but the edges that make up those faces are on Default.
This may not be causing any issue in this particular model, but when you start working with large complex models and you have your edges on a different layer to the faces I can see where one movement of a face without it’s edges can have disastrous consequences.

1 Like