[Discussion] To update blender or to not?

Over the course of large projects, blender updates at least a couple times. Over my DEEP Space project, blender went from 2.51 to 2.63 (from memory).
The project eventually died when I went to university and ran out of time, but before then, it was on it’s way out anyway due to a few things.

One of the large issues encountered in DEEP Space was that of blender updating. When blender changed from 2.56 to 2.57, a whole bunch of texturing things broke, other changes in blender’s internal file-structure required me to open, save and close all the project files on several occasions.
I am aware that the time over which I made DEEP Space was one where a lot of changes happened in blender, and generally, support for older files is better now.

So I guess my question is this:
When you start a big project, do you ‘freeze’ the version of blender you use? Or do you, every two months, spend a couple days updating things?

I update blender every two months to get new features for blender and the blender game engine.Then i save my openworld videogame in it.Especially if i see something in resources i want.I just use logic bricks and python code from resources i can find.

It depends how much the project is Blender dependent and how complex it is (animation involved? rendering?). For projects where I use blender just for modeling I like to use the most recent version, for complex projects it is too dangerous to update. Fortunately it is easy to use multiple Blender versions at the same time. I just need to make a note which version was used for which project. Sometimes when the project is in an early stage it is worth updating and then trying to repair what’s broken.

Before I upgrade, I always copy my “old” version into it’s own folder. That way, if, for some reason, I ever need to “backtrack” with my versions, I still have access to it. I keep it in the “Blender Foundation” folder, but rename the copy so that I know which version it is. So I’m currently running 2.71. When 2.72 comes out, I’ll copy the current version into a “Blender 2.71” folder. Note that I COPY, not move. This is so that, theoretically, I should be able to keep my preferences when I upgrade. (That doesn’t always work the way I thought it does, or thought I heard that it’s supposed to, so I always do a quick check to see what I need to “reset”.)

Still on 2.49b I even downgraded code from higher versions. 2.49b has a more pleasant interface. I come from years of professionally using 3dsmax and I hate those kind of little windows with tabs everywhere workflows. I found 2.49b game runs signifivantly faster too.

Unless AA is lag free and you can push much more visual data, I won’t update. I think 2.49b has all the tools I need. And it has more keyboard shortcuts, gestures, etc. I don’t have to think about the interface.

Real time shadows being slow is the only regret I have.
Another con is a lot of resources online disapearing.

I wanted to add, I am usually against most updates, I stuck with Photoshop 6 for many years, now it’s CS4 because windows cannot run 6 anymore and full screen mode has become annoying to work with after CS4: wild auto scrolling, no more access to menu bar if in full screen, streaks if you draw all the way to screen border. Lots of updates screw up functionality and it’s pointless to complain, so I usually refuse to upgrade. Polygonal lasso in pshop has been laggy and crash inducing since 7…

As for me, I will try to modulate as much as possible. Separating initialization and logic from the beginning. I use lots of variables and comments and expecting changes to the framework. I will try to cope up as much updates as possible. I do keep multiple version from the beginning of development to the latest version.

LOD and the recent armature threading are probably worth any pain upgrading would cause,

I update whenever there’s a new update. I don’t just use the bge, I like using cycles, and there’s almost no recent update without a cycles improvement.
Sure there are compatibility issues but I personally don’t mind any of them, as they’re nothing compared to the advantages of updating.

Still on 2.49b I even downgraded code from higher versions. 2.49b has a more pleasant interface. . . . and I hate those kind of little windows with tabs everywhere workflows. I found 2.49b game runs significantly faster too.

Unless AA is lag free and you can push much more visual data, I won’t update. I think 2.49b has all the tools I need. And it has more keyboard shortcuts, gestures, etc. I don’t have to think about the interface.

Real time shadows being slow is the only regret I have.
Another con is a lot of resources online disappearing.

X 2

I wouldn’t consider myself a game developer. I mostly make games for my children, grandchildren, and I to play. None are really good enough to sell.
But we have fun, and I have fun making them, and that’s what matters to me.

So I stick with the build I started with. Mostly 2.49b, or occasionally 2.66
If I need anything fancier then I’ll use Free Unity3D.

To me, it seems a waste of time trying to update python scripts, redoing animation etc. when I could be working towards finishing the game (As if its ever finished)
Just my .02.

That being said, I do like the new features on the newer builds, Especially the Lib-Load, Lib-Free . But the interface is painful for me. Plus, I just don’t have a PC that will run the newer builds without crashing.

I update pretty often, but I also keep older versions of blender installed on my computer if I ever needed to go back. So far, I have never updated and lost a ton of work.

A big thing I would think to play a factor in major chaos would be a major update. Jumping from like .63 to .71 would run more risk for large issues than updating every update or so and fixing the small things that may happen along the way.

Plus blender is adding cool/useful stuff every update, and often times the possible gains > risk of something breaking.

@sdfgeoff:
Is there a feature in the new version that you want to use? If not, then I’d say don’t bother updating.

A year and a half ago we updated from 2.45 to 2.59 and it was a nightmare. We did it for Python3, though I still sometimes second guess if it was worth it all. Updating gave us a whole new set of (Blender) bugs and compatibility issues, particularly with older cheaper hardware. 2.45 -just worked-. I miss that :frowning: Better the devil you know, if you can put up with it.

It never really concerned or bite me until 2.70. Now I would say stay with the version the project was created in.

Of the more recent versions, the one that seemed to give the least problems with BGE games is Blender 2.69 (it’s a lot like 2.49 in this respect but with more features).

Though the upcoming version is pretty close though thanks in part to that 3 month contract given to Moguri, my projects seem to work okay at least (or works again with a small amount of work).

wrectified runs smooth 60 fps on 2.70 its complicated as all heck,
without recording if I uncap the frame rate I get about 110 fps

Ideally, you could update everything, and your game wouldn’t break. However, we don’t live in that ideal world, and when things break, it would probably be a good idea to “freeze” the blender version for that specific project … unless the break can be resolved with a few trivial changes, in which case, I would simply implement those.

also, us analyzing what breaks, is how we debug new updates,

I noticed the armature not running instantly, on 2.71 because my projects use

always------and------run armature

a lot.

we are all different so we might each expose something else in the new version,

thus ensuring kickassedness as well as getting new features,

just back up your stuff,

in a new version worst I have had to do,

take broken object
add new object

Join old to new

delete new model faces

add in logic again,

fixed.

one time I had to do flip normals on like 50 planes…

Since Blender is open-source, it’s more a question of how often you want to update. As others said, most update when they need a new feature or important bug fix. However, if you update so frequently, then you’ll have to frequently update the code also. Personally, as a developer, I pretty much update several times a month :eyebrowlift2:

However, I still don’t get why people stick with 2.49 since there have been so many improvements since then.

It’s a quality software to me, it has everything I want. I can nearly do anything simply typing on the keyboard. I have found a workaround to get what I need in game every time something seemed impossible.

More importantly, I left the babeltower interface “designs” of 3dsmax and Maya behind for blender, I am not going back to working with a soft full of little boxes you have to scroll through and go click click click with multiple prompts. I am learning Solidworks right now and it’s a freaking nightmare of sub menus, tabs and prompts, it’s impossible to work instinctively in it, I haaaaaaaate it. That’s how all softwares are being made these days including the post 2.49 blender versions. I am making my game as a hobby for fun, so I will pass on that frustration and stick to 2.49b.

Updating or not depends on what you want to achieve.

You need to ask yourself

  • do I really need the update? There is no need to update, if the current version is fine as it is.
  • do I have time to spend on BGE update? It will cost you time 100% sure.

Basically the BGE you start with is part of your project. All future work will rely on it.

If you update you might get incompatibilities to your current state of the project. This is a known risk. You can ease the situation by
a) performing sufficient test,
b) ensuring to be able to role back and
c) to upgrade without changing the project.

A) is something a lot of developer like to forget. If you can’t test and repeat the test … Do not update as you can’t be sure the update broke something. Indeed you cab start testing later.
B) is pretty simple if you use a versioning system
C) this is an essential step at the development: do the upgrade, fix the problems you got ehile upgrading. If you add new features at the same time … You can’t tell if they are caused by the upgrade or by the new feature. And you can’t roll back without loosing the new feature.

Idealy the upgrade runs smooth without side effects. Without tests you can’t be sure.