First, sorry for those who may have read this allready somwhere else, as we posted in several places.
So, we are a rigging team, here at Mac Guff Ligne, and we are moving forward to add Pyhton to our rigging toolSet (finaly!). Even if we are beginner in Python, we understand the benefits of it (its object oriented nature, the vast amount of available libraries, large community, etc...), and we know that the standard implementation of it in maya should/could have been much more deep and powerful, as Pymel seems to prove. We read a lot about it everywhere, and it seems that pymel is a very powerful example of what the implementation could have been, both on object and api sides.
We are truly enthousiasts about using pymel but our R&D department is almost against the use of pymel.
Their arguments are:
Probable issues and the support it might need.
As this projet doesn't belong to Autodesk si it may stop at any time.
Major changes in pymel future versions may occur and obsolete large portions of our code
As it's another layer of code, it brings its own bugs.
OpenSource
Our arguments are (based on various readings over the net): - The ObjectOriented implementation rocks. - Smaller and clearer code, thus faster development. - Smarter Api integration. - OpenSource (^^)
So at last, we decide to turn to the community in order to gather the maximum pros and cons. We'll be glad to hear from everyone (Pymel developpers opinion, of course, is a must), for any advice, known issues, problem with support/continuity, knowledge and experience you have and of course everything you can think about it!
Its both a benefit and a downside ate the same time. To illustrate the point its like a pizza place that offers you 30,000 toppings versus the one that offers 5. Well what if you like your pizza with 5 your still paying for the 29,995 other offerings in way.
and you really only want to ever use one or 2. Its very easy to go and reprogram a program just because you had the libraries. And didnt happen to know the ones you do have. So knowing if you swap a good thing for something worse gets harder to know.
well same can be said of all code. Even autodesks. they dont actually guarantee maya will stay on forever they too may end a moments notice. Worse you paid for the support whan they screwed you over.
I would say pymel is much safer than what autodesk gives you. You can just not update when pymel updates. And you can read what the schanges are fixing a few lines of code is nowhere near as much as what autodesk puts you up with right now.
Sometimes it does sometimes it doesnt. A lambda notation language rocks, but somehow magically even if its the defitively best programming paradigm weve come up so far it sucks terribly much.
Object orientation is just a good thing if the problam is easily described in objects. It also makes for a lot of retyping when things dont go that way. Tough puthon duck typing handles this quite well.
so for python the issue is:
PRO:
its powerfull in teh right hands.
CON:
its a powerful way to shoot yourself in the leg when you dont have the right hands.
For pymel. not much to say its just a layer of abstraction more it makes some things baeutifully fluid and saves you the time doing it but than again when you get to it maybe you should.
PS: theres really no sane reason to convert over, just do the new stuff you can easily do the way you do now the way you do now and the stuff you cant with python, including pymel if you wish.