Just out of curiosity do you only use debug notifications/traces to rely on whether scripts are working or not? Stack profiling will tell you a lot more about whats ACTUALLY going on rather than what you expect to be going on. I've had scripts outright fail because of other scripts that were completely unrelated, until I threw up a stack profile and saw one particular script was stuck in an infinite loop as the stack profile continuously grew in size. Debug traces won't provide you with this level of information.

Also, you may want to start doing more complicated edits on a vanilla version of Skyrim if you are not already (No mods except for the mod your making).

OK, I looked at this and... you probably aren't gunna like my results. Everything worked completely normal and perfectly for me.

Following your instructions specifically, I DID come across a glitch in the CK that may have caused PART of your problem. Back in early December I wrote my Guide to preCK Moddery (avail on the Nex)... in it I explain what one has to do after creating a custom WORLDSPACE. This bug seems to have persisted through the Garden of Eden straight into the plain/new CK. Here's the gist of it:

CK does NOT create a LAND record for any custom worldSpaces unless User manually edits the terrain of EACH cell using the landscapeEditor. This can be done using a 1 or 2 radius rise or fall of the slightest magnitude. You can even change it back (manually... NOT by using the 'back/undo' function). Snipping over another cell's LAND record DOES work, but it duplicates that cell's data (so probably won't fit your desired motif).

If you DON'T do this, upon entering the custom worldSpace Player instantly DIES... using the console to resurrect causes Player to die the Perpetual Death (it just keeps dying and dying... quite the moving scenario). This is similar to what you describe as "levitating". As soon as the initial cell's LAND record is written, the cell operates as normal; but all surrounding cells seem to be WAY down low. If you try to walk off the first cell, you fall to your death... unless you move Player back across the cell-border before it hits the bottom.

I dunno if this is what caused your issue, but it's the only thing remotely close that I could find. I DID find that when placing navMeshed statics in the worldSpace, sometimes Player would seem to be floating way up high, but still dies... similar to the levitation. I never saw disappearing statics or anything of the like.

Anyways, following all else to a tee, it works perfectly... didn't even have the NavMesh Bug. As soon as I converted the ESP to an ESM, it causes a form of NavMesh Bug (I call it NavMesh Bug 2, or NB2)... works fine for FirstArrival, after fast-trav-return all actors with inter-cell AI are broken. I call this NB2 because with this one the actors disappear completely, whereas plain ole NB1 the actors only 'Wander' to elsewhere in the cell (or to an adjacent cell). I explain this better in the NavMesh Bug thread.

Splitting the ESM into an ESP/ESM combo only worked when ALL Vanilla navMesh data was in the ESP. I simply Snipped over the entire NAVI grup and the entire Tamriel WRLD grup... all else remained in the ESM. Worked fine (again). Then I added a door to the worldSpace and did some other testing; but never found anything like you're describing.

Sometimes when I use Snip, the records look like they've been deleted or aren't there. I don't know what causes this, but they are NOT deleted... simply not displayed. Exiting Snip then restarting usually shows them correctly again. Like the CK (and specifically navMeshery), I keep Snip sessions to a bare minimum and restart frequently (to avoid session-specific drama, such as inevitable garbage collection and what that can do to data in memory/being saved).

I dunno if that's the case with your mod; I know you're a savvy user, so I doubt it. But if your mod was THAT severely damaged, as seen after loading in the CK, it sounds like you have system-related issues. As in, your CK may be corrupt, Snip may have been used too long during that session (before a save caused the damage), or you may have a virus or some other corruption issues. You've tried 'verify content' through Steam to eliminate that possibility? If it were virus-related, most scanners do NOT pick up everything... so you could have something for months until a scanner's update finally catches it. Stupid punks are getting VERY crafty as to where and how virus are transmitted and what they affect.

But I still have yet to see ANYTHING damaged by Snip... and I've been using it since TES4Snip preCK. I've done a TON of splice jobs, altering data, this that & the other thing. I know it decompresses data, but that's fixed when the plugin is loaded in the CK and re-saved.. as is the record count and file header (if CK changed it; which it does if it auto-corrects the record count).

I practice strict version control as well; if myMod.esp is in the CK... after I save it, I rename the ESP to myMod1.esp. Without re-loading the plugin, I continue making changes... after saving again I name rename the myMod.esp file (which CK creates fresh) to mymod2.esp. Then, whenever I use Snip to alter one of those, I make the changes and save the file to mymod2a.esp then myMod2b.esp etc. I either note in the actual filename what the change is (myMod2b LAND-splice.esp), or write it in one of my 'notes' text files (Notepad++ is ALWAYS open on my system). This way I always know what causes what, how far to restore something to 'fix' drama, etc etc. I assume you're already familiar with this kind of stuff, so this is more for the benefit of other people reading this... heheh.

I also agree with you about testing mods using clean and 'dirty' saveGames; as opposed to ALWAYS coc'ing from the Title Screen. I use coc for MOST of the testing, as it's sooo much quicker. But once the mod is at a presentable level, I start testing with saveGames to ensure everything is the same as other Users would experience.

Also, I found that onCellAttach and onCellLoad fire simultaneously... my tests placed notifs in both events (both in the worldSpace and the Vanilla cell); as well as before the if-endif, inside it, and after it. I did find that Load events don't necessarily fire when you'd expect them to... less in fact. But all notifs from both events were sometimes intertwined in their appearance and sometimes one event's before the others.

I've been using TesVsnip on worldspace ESM's and ESP's for months (for MERP) and never encountered any corruption of data. Sounds to me like it's still a mystery what exactly went wrong, too early to blame TesVsnip in my opinion.

Gecko for Skyrim is in the works, so it hopefully won't be too long before something that's actually reliable AND built properly for Skyrim is available.

As for flipping the ESM flag, Wrye Bash can do this safely for Skyrim.

If you're using Snip, you're playing with fire because it cannot process the compressed record structures properly.

im not going to rule out the possibility that the cause of this strange issue could be a freak occurance (as i too have been using snip to cut/paste/edit/rename/move/ thousands of records successfully until now)

i can't conclusively prove it, because i cannot replicate the glitch on a new mod either. i am almost half way done rebuilding the mod and everything works perfectly fine right now. i even tried re-snipping just for test purposes on a duplicate copy and that copy works fine as well.

although just to be on the safe side, i am not ever going to use snip, since just about everything i would normally use it for can be done in the ck anyway. right now im only using snip (read-only) as a convenient way to look through my records in a nice and structured way. other than that i dont feel like gambling with it any more.

This line of thought is upsetting because the CK won't allow you to remove the ever-present dirty WRLD record, which, I've noticed by comparison of (the same) mods with this record and without it, seems to cause CTDs on exit to the main menu.

This record is only added if you save with Update.esm active, but it really shouldn't be doing that at all.

Which record is that?

are you sure?

i am able to remove the entire WRLD group using the CK (my current mod is also dependent on update esm). also never crashes when exiting to main menu.

the whole ignore thing in the details menu is tricky at best, took me a while to get the hang of it, but you need to flag ignore, and press ok to re-load your files, then save and close (if you want to delete a whole group, you also falg ignore on all the cells it contains as well as references in the cell just to be sure). this always works for me. never had a "sticky" record that refuses to be deleted using this method.


i have had the crash to main menu in the past while working on other files, but it was always on a WIP esp (never on a finished esm), but not related to the WRLD records, the problem eventually went away on its own by the time i finished working on the file (on some of the cases), or was close to finishing. most of those cases where it wasnt ever solved was because the esp went past the "memory barrier" for interior cells and would CTD regardless if it was main menu or another worldspace.