My one big concern about this that users find a way to exploit this behavior for themselves.  If it's too easy for users to create tx they know will get stuck and expire, it's no different than letting them cancel their zero-conf transactions.  i.e. I pay 0.5 BTC in a store for a candy bar, so I send it using a combination of inputs and fees that I know will lead to it being stuck and expire.  

On the other hand, if such conditions are deterministic enough, it could be detected by the recipient and flagged.

It's not a huge deal, but it's something to consider.  

-Alan



On Thu, Apr 12, 2012 at 2:38 PM, Jeff Garzik <jgarzik@exmulti.com> wrote:
Not sure whether this rises to the level of a BIP or not, as it is
largely an implementation change.

One of my From-Day-One complaints about bitcoin is that transactions
behavior could be far more deterministic (predictable), from a user
standpoint.  Transactions in the current system can easily remain in
limbo forever.

One big step in making transactions behave more predictably would be
to remove transactions from the memory pool, if they have not made it
into a block for a couple days.  i.e.

1.  N = 1 or 2 or whatever the community prefers.  Ideally enough time
for a third-tier miner, mining strange TXs, finds a block.
2.  H1 = height of block chain, when a TX is received
3.  H2 = H1 + (144 * N)
4.  If block chain height reaches H2, and TX has not made it into a
block, drop TX from memory pool

Although this only impacts a small amount of TX's ultimately, what it
does do is give us the ability -- once miners have upgraded to this
rule -- to tell bitcoin users that their transactions "expire" after N
days.

Backwards compatibility should not be an issue; clients are free to
retransmit their TX at any time, as usual, thereby "resetting the
clock" for all peers who have forgotten the TX in question.

Once in place, clients may then implement code that notices a TX has
expired (read: likely to have been forgotten by the network, assuming
they themselves have stopped retransmitting it).  Then you can start
working on wallet/coin recovery, perhaps resending with a higher fee
etc.

The above change is not really "fill-or-kill" but it should be a big
step, opening the door to deterministic TX behavior.

--
Jeff Garzik
exMULTI, Inc.
jgarzik@exmulti.com

------------------------------------------------------------------------------
For Developers, A Lot Can Happen In A Second.
Boundary is the first to Know...and Tell You.
Monitor Your Applications in Ultra-Fine Resolution. Try it FREE!
http://p.sf.net/sfu/Boundary-d2dvs2
_______________________________________________
Bitcoin-development mailing list
Bitcoin-development@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bitcoin-development