The other day I was listening to Mahler's 9th Symphony. An old college friend of mine, David Handel, used to heap scorn on the piece ("Oh. That thing..."), arguing that Mahler died before he could give it final revisions and authorizations, and that it was therefore not valid. Now, the musicological discussion of Mahler's revisioning process is not a battle I'm well-armed to fight, or even particularly interested in, and his 9th remains one of my favorite pieces.
It does raise the question though, of when is a work of art finished... how do you know? We deal more or less with abstractions, in the sense that we're extracting some kind of essense and mapping it onto something else. We look at a tree and then put an image of the tree on a canvas. There might be a lot of feedback (does it look like a tree?), but it will NEVER be a perfect copy of the tree, nor should it... slavish copies tend not to make for interesting art. You are therefore NOT exclusively determining "done-ness" based on verisimilitude to the tree.
Some artists seem to not have a problem with this, others clearly struggle. William Bouguereau, the great 19th century French master, churned out finished painting after finished painting in a seemingly unending flow of canvas. I doubt he really struggled with this much at all. Da Vinci, on the other hand, had a relatively small output, and constantly revised his works. He kept a number of paintings with him all his life, and apparently was still working on them when he died. On the musical side, Schubert and Mozart both wrote reams of music in very short lives, while somebody like Bruckner lived twice as long and probably published fewer than 25 or so pieces, revising them constantly. There are multiple versions of many of his pieces, and to the glee of musicologists, there is a ferocious debate about which version is "Final". The problem makes terrific dissertation fodder for aspiring academics.
I often wonder about the significance of the creator's stamp of approval. There are plenty of works that the artist considered "finished" which I think are perfectly awful, then there are pieces like Mahler's 10th, which is an even more extreme case than his 9th. He left it in rudimentary sketch form -- highly fragmentary and incomplete, yet 60 years after his death somebody made a "performing version" of the piece by stitching the existing notes together with newly composed material. Is it Mahler? Is it not Mahler? Does it matter? Whatever the case, it is a tantalizing glimpse into what it might have been. And it's very, very beautiful music... I love it beginning to end.
From my own experience, I guess I'm one of the luckier ones. I don't find myself obsessing over my art. I almost never rework older paintings. My feeling tends to be that if the current one didn't go so well, then I will do better with the next one.
As far as knowing the moment of "doneness", as a painting approaches completion, there does become a sense of tension between the temptation to keep trying to improve it, and the fear of over-working. I've all too often gone too far, and my well-intentioned "fixes" ended up botching the job. I guess part of growing artistic maturity is learning to respect that sense of completion, and go no further. Lately I've taken to enforcing that respect by scraping all the paint off my palette when I feel the moment has arrived.