[pavel59]

Yesterday I started experiencing this strange error on my 2.7 release, both on 32 and 64 bit versions for Rhino.
The same file works well, then suddenly it exports to MXW render and it shows this message: cannot read MXS file... No error. Then it quits.
I removed an object from th scene and it worked again
Today I had the same issue with another file, tried on both versions, changed parameters, no way.
Then I uninstalled rel 2.7 and reinstalled 2.6, now the generated MXS is 318MB against the 108MB of the 2.7 but it works flawlessly.

In the past I'd been working with MXS files of 1,5GB and above, on the same machine, never ad an issue.
This must definitely be a bug in the new file system.

Paolo

[JDHill]

I am aware of no issue like this, and I just checked by writing an MXS of 560MB. You mention that removing an object affected it; are you able to reproduce the issue from scratch, or is it dependent upon this object? If, on a failing export, you check the Log Viewer window, are any related errors reported?

[pavel59]

no. it happened on two completely different files and setup.
nothing in common to each other, not even materials.

And I was able to repeat the same on both 32 and 64 bit.

No errors on the log.
It just say
Unable to read the MXS file. No error.
And then it shuts down.

Trying to open the MXS file in Studio hangs up Studio.

The model which I added (an offshore catamaran) rendered successfully alone.
When imported on the shadow vessel it generates the error. Hiding the cat and it renders.

The other model is a large schooner (a classic sailing boat) 38m in length (the shadow vessel is 90m) and it doesn't render at all.
Reinstalling V 2.6 everything works fine. All models, all configurations.

My system is a 8 cores Dual Xeon, 32GB RAM ECC, Win7 Professional, Quadro board.
I rendered successfully much larger models.

Paolo

[JDHill]

The point is that this is abnormal -- it is not something I am going to be able to figure out by theory alone. If you are not able to reproduce it from scratch, then I would need to inspect the model, or at least the MXS file that cannot be read.

It sounds like something that comes down to a particular piece of geometry, or geometry produced by a particular method -- I cannot guarantee that something which is valid in 2.6 would also be valid in 2.7. In theory it should be, or some kind of error should be indicated, but I can't say much about that, since it is not something under my personal control. It could be anything -- an unknown bug related to a particular image format used in the model -- anything, really.

The best thing would be if you could upload the models necessary to reproduce the behavior. Either to a file sharing site, or if security is an issue, I can give you a FTP login to upload to my own server.

[pavel59]

Hi, JDHill
sorry for not replying earlier but I wanted to find at least oen file of those which didn't render.
I have one and I cand send to you. Can yoy send me an email with instructions how to delivery the file to you?

In the meanwhile I've been privately contacted by other users having the same issue. I invited to share on this thread.
I hope my file will help you to find the bug.
I'm not in a hurry at the moment, I can survive with 2.6


Thanks

Paolo

[JDHill]

I corresponded with someone else experiencing this issue today, as well. In that case, the customer sent me files via dropbox, including the source .3dm, in which I was able to find four bad meshes (degenerate faces, meshes not visible in Rhino viewport), by method of trial and error. So, I have provided the engine developers with a simplification of this scene, and it will not be necessary for you to send me yours, unless you believe it shows a different error. If you do want to send it, my email is jeremy at nextlimit dotcom (dropbox worked nice for this).

[JDHill]

Please check this again in 2.7.10 -- these bad meshes should now be tolerated.