Should one post a paper on the arXiv if it is not intended to be published? A brief description: I have written a paper which contains a new result which I believe is somewhat important but not vital to the field.  It is a generalization of an existing proof to get significant new information, in a framework that did not exist at the time of the original paper (not by me).  I do not believe the result can be recovered from the original result, only from the proof, which is lengthy.
The paper as it stands is self-contained but a significant portion is a reproduction of the original paper (recovering additional information from various lemmas).  Some lemmas are new but many are old.  A copy of it which does not reproduce original work is only about 5 pages long but is very difficult to read.
It is my adviser's opinion (and I agree) that neither form of the paper is suitable for publication, for different reasons.  We both believe the result on its own is significant enough, though.  So, the question: would posting the long form on the arXiv as is be appropriate?
To head off the obvious question: yes, the author of the original paper is prominently pointed out in the abstract and the introduction, where it is pointed out that we follow the original proof closely in most cases.
 A: I've seen this proposed model for the future of publication: The mathematician puts up her paper on arXiv (subject perhaps to some minimal anti-spam safeguards) and then "submission" consists of asking a "journal" to referee and endorse the paper. The journal consists of pointers to articles which have been accepted for review and found to meet the standards and conditions of that journal.
But for now, for what it is worth, it seems quite reasonable to me to submit to arXiv, as long as your advisor agrees. Whatever the field, I'm sure you could find some journal which would be happy to print the paper (if valid and well written) along with submitting it to the arXiv. But why do that, I'd think, if it is not going to bring it to the attention of any readers?
A: Based on what you say, your paper would be valuable and useful for the mathematical community. So I think you should put it on the arXiv, with the remark in the comment field that the paper is not intended for publication.
Update. I meant "the paper is not intended for publication in a journal". Thanks for the comments and the (unexpectedly) large number of upvotes!
A: Given that arxiv is a pre-print publication site and many of the hosted papers are not published elsewhere, yes i would say you should post it there.
Note that arxiv (and various similar sites like vixra) is good for other functions as well. For example we know the journals are sometimes heavily refereed and not everything published might be the best (for example if by an otherwise important author) and/or not everything important is published (for example by a relatively unknown author or sth very contrived)
Apart from that, personaly i think that sites like arxiv enable research and technology to go further by allowing partial or otherwise un-published results (see above) be published even in this format
i make heavy use of arxiv and of course i am aware that sometimes invalid results get publishd there.
Furthermore this would give you something like a copyright license (e.g "i published it on arxiv on that date") or to be more precise since i am not fan of copyright. It would give you the chronological advantage if you like
A: The answer to the question "would posting the long form on the arXiv as is be appropriate?" is yes, as others have said.  But there are other questions you should ask yourself.


*

*Can you simplify the proof?  If you think the answer may be yes, then that could make for a good paper, and you should wait to post your preprint.

*Do you want to quote this result in other work you do, or have others quote it?  Then, in my opinion, it is better to have it in a peer-reviewed venue.

*It sounds like you are a young mathematician, since you mentioned an adviser.  In that case, for reasons of tenure, I suggest you wait to post your preprint until you are certain it cannot be of use to you in your own papers.  [This advice may not apply to all research areas.  Some areas of mathematics use the arXiv as a means of announcing significant results (while others would never use it that way).  So use your own discretion here.]

*I'm a little surprised that you are of the opinion that both versions of the paper are not suitable for publication.  Are you sure you couldn't add/subtract something from one of the versions, which would make it of sufficient interest to others?
