Let me explain. PyFFI has an option to create patches directly (--diff), and also an option to update nifs using patch files (--patch) - actually this was supposed to address the issues with the original patcher batch files. But I never really documented it properly, and never included xdiff or similar. Also, it only uses in/out (confusingly) - an extra "patches" folder as you suggest, would be much better.

So I think I'll upgrade this feature to do essentially what your batch scripts are doing (but it wouldn't rely on forfiles.exe and so on).



Yes, will do!!

Aside, as it looks now, these updates won't be for 2.1.8, but more likely for 2.1.9, as I'm compelled to ship 2.1.8 already this weekend to release some longstanding bug fixes for the blender nif scripts, which require an updated pyffi.

can someone explain to me how to repack the BSA. When I did it the first time, and it crashed the game, I used "Add Folders" and selected the toaster/out folder. Should I be using the "Add Files" option, and selecting all of the 1000's of meshes?

If I should use the "Add Folders" , how "deep" into the folders should I go? (toaster/out) or (toaster/out/meshes)

Also, will it actually screw anything up if I move the files from the toaster/out folder to say, a new folder on my desktop?

You don't want to add just the out folder, PyFFI only puts the meshes it modified into that folder, not the ones it skipped. This might also explain your missing mesh trouble in Nehrim.

When repacking the BSA in OBMM, the folder you want to select is one folder above the meshes folder (so select the in folder and if it contains the meshes folder, the meshes will be properly packed).

Don't forget to re-pack your trees and DistantLOD (if not using TES4LODGen) folders either.

Alright, I managed to get my Nehrim Meshes PyFFied! Everything appears to be working nicely...for now at least.

So, when I Pyffi the oblivion meshes, do I have still have to worry about those http://sites.google.com/site/oblivionpoinfo/optimization/pyffi that have the added space?

Only if you're on Vista/7. Even then, just don't touch the folders after PyFFI is finished with them and you won't have to deal with the leading spaces being truncated.

EDIT: In other news, the OP overhaul has begun it's trial run. If anyone has any questions or comments or additions that could be made please don't hesitate to ask.

What's the latest version of PyFFI I can use with Python 2.6.5?

Well, I followed everything to the letter, and after Pyffi'ing the Oblivion Meshes, my game is crashing when it loads a save. Just like before.

You have any idea what the problem could be?

I reset the timestamps, so the Meshes.BSA says 1/1/2006. I copied the Meshes.BSA from the Oblivion\Data directory to the Nehrim\Data directory. Could it be because Nehrim pulled the original Meshes out from the Oblivion\Data, and since I've replaced them, it's causing it to crash?

I personally just manually deleted the leading spaces from those meshes (using XP) as Arthmoor's UOP updates modify those cave objects to look for the meshes that don't have the leading spaces.

If you're doing any modding at all, you'll want to try and keep those leading space files from getting deleted. The CS will not recognize that the UOPS has fixed this. Unpacking your BSA somewhere, doing PyFFI on them there, then repacking them will work. Just don't copy/paste them or you'll lose those. Again, this should only be an issue with Vista/Win 7 64bit.

New beta version (2.1.9 beta1) including freshly written recursive patch scripts is available:

https://github.com/amorilia/pyffi/downloads

Thanks to OnmyojiOmn and jonwd7 for their initial patcher scripts - I did not use any of their code (since the new ones are written in python, not in batch) but my code is obviously inspired by them.

Beware: use the new patch scripts only for testing - they are not suitable for "real life" use until more properly tested. For now, stick with jonwd7's patcher if you want to do real life patching, and not just testing. Aside, note that the patch_recursive_apply script will copy files from in to out if no patch file is present, so that should mitigate any issues with missing files, at least when using patches. Probably I should provide an option to do something similar with the other scripts.

The patches use xdelta3 (version z), which is now also bundled with the windows installer, so earlier patch sets (including those just uploaded recently by jonwd7) should in theory keep working, although I haven't actually tested this yet.

Quick instructions:

• to create patch sets: copy originals into in/, run optimizer so updated are in out/, run patch_recursive_make.bat - the patch/ folder will fill with patches
  
• to apply a patch set: copy originals into in/, unpack patch set into patch/, and run patch_recursive_apply.bat - the out/ folder will fill with updated files

If you find any issues, please report them on the tracker as usual:

http://sourceforge.net/tracker/?group_id=199269

I used the latest xdelta as well, at the highest compression level. I'm not sure what the size difference is between midway and maximum, but I needed to get the vanilla mesh patches as small as possible. And there was a pretty big difference between the min and max compression. It's unfortunate Windows does not have built-in Python support (*cough* OS X *cough*), because Python is my preferred language. I still want a patching solution free of additional installers and dependencies, but not as primitive as a batch file. Though if I don't come up with an executable at some point, I'd rather point people to patching with PyFFI itself, for the people who don't mind having to install Python and PyFFI to do it.

The OP says "Python 2.6.6 (required to Run PyFFI)". But when I go to the link for documentation and look at the requirements, I don't see anything about Python 2.6.6.
I also looked at the readme.rst for 2.1.5, 2.1.6 and 2.1.7 and saw no mention of this requirement.

So I'm still looking for an answer to my original question.
What's the latest version of PyFFI I can use with Python 2.6.5? Thanks in advance.

Python 2.6.6 will work fine. That's what I'm using. Just be sure you don't use the 64bit Python.

I'm using whatever comes with the Python package for Wrye Bash found on Nexus. I think it's 2.6.5 -- it may be 2.6.6 though.

It was 2.6.5. Thanks!

Ok, after gaining a little confidence after pyffing all the meshes in CNRP with a blank skiplist and not getting any freakiness, I decided to go all out and pyffied my Oblivion - meshes.bsa without a skiplist.

2 hours and counting ... nothing abnormal W00t PyFFI!

Why on Earth would you do this? There are skip lists for a reason...

Hey just did my Oblivion - Meshes BSA, 11 hours!

Are there any areas and items I can go and check that have known issues? Im seeing that helmets and the roots and stuff on the first page, but is there anything else(specific dungeons or towns or whatnot)?

Just "coc testinghall" in the console. Go to the creature grove and check out the creatures. Go to the other rooms and check out all the clothing and clutter.

Hmmm, I'm using the default (no extra settings), is that maximal compression?

(EDIT: ok, just did some testing, and it isn't; I'll update the next beta release to create patches with maximal compression)



Excellent point - there are two types of users:

• Those who want to *make* patches will want pyffi anyway and it is easy for them not to need an extra tool.
  
• Those who want to *apply* patches. For these users, installing the whole toolchain is complete overkill, and having a simple batch file that does the trick is great (easy to maintain, and it's open source by definition ).

Wonderful! Be sure to check that mothernightstatuebase.nif, people keep reporting that this doesn't work (I've never actually observed the problem myself, but maybe I don't play the right factions... ). If you (or anyone) sees an error, and can track down the cause down to nif level, then that would be absolutely awesome.

That's great and all that he played a whole two hours with no issue, but those skip lists have been built up over many years of testing. Two hours of testing shouldn't negate the years of multiple input from other people who have had actual problems.

I would think about this person differently if he/she PyFFIed everything and then intentionally encountered the problem NIFs that everyone reports, in some kind of scientific manner. (Not just running around the game world, probably NOT encountering the problem NIFs)

I know PyFFI continues to improve, but I don't see why somebody should ignore a skip list. It's not like PyFFIing those things will bring any additional improvements.



No, you have to use -9 in place of whatever the letter is for the patch flag (-e maybe?)... It takes a bit longer to make patches this way, but not to apply them. You may want to try a middle-of-the-road setting like -5.

My friend, I'd like to remind you that:
1) I'm not encouraging anyone to do this
2) I've seen that amorilla immediately fixes any specific problems reported to the sourceforge tracker

I do have my old (skiplisted) pyffi 2.1.7 files on backup - just in case. But I also DO intend to report anything specifically nasty that comes up in my un-skiplisted meshpack. Afaict, amorilla has repeatedly said that its perfectly safe to run pyffi wothout the skiplist provided any anomalies are reported - which I DO intend to follow

Thank you for your time!

Thanks, I also just found out about that, and the update will be included in the next beta release.

Indeed, Gratis_monsta is one of the regular (and well appreciated!!!) testers - but uses another alias on sourceforge whence perhaps the cause of confusion. Anyway, the skip list serves two purposes really:

• To prevent optimizations to go bonkers if run on mods that have nifs with egm or tri files in vanilla, but that don't ship it (so pyffi doesn't see it when optimizing that it cannot change the vertex ordering in these nifs).
  
• To prevent optimizations to go bonkers on any other nifs.

In an ideal world, there wouldn't be any nifs in the latter category, so, in that world, you can safely pyffi without skip list, provided you are running it on a vanilla unpacked bsa. Of course, the world isn't ideal (yet).

Thanks, Ill check it out.

Took a quick look over Bravil's Lucky Old Lady and there seems to be nothing wrong with her base Some pics (cropped for bandwidth)
http://img163.imageshack.us/i/screenshot13mt.png/
http://img822.imageshack.us/i/screenshot14p.png/

I double checked with FormID finder to confirm that the mesh wasn't a different one (I have Blood&Mud installed)

Also took jonwd7's tip and "coc testinghall"ed and tried out some items - nothing extraordinary. Went to the creature farm and had a HUGE melee (maybe MMM/FCOM at work there) but nothing special Also confirmed that the rat underbelly seam is now almost gone (only shows up when you have the rat close to you and have a torch in your hand)