Market-orientated transaction pricing

[gavofyork]

Avoid a strict gas-limit and let the market decide through using
a priority queue based around gas pricing for transactions. In
periods of low transaction volume, they'll be processed for a lower
fee.

[tomusdrw]

I would move this to line #334 - cause it will happen only in that case.

[gavofyork]

would like to, but hashes has captured an &mut of into in this scope so it has to escape before it can be removed.

[tomusdrw]

Maybe explicit drop(hashes) would work?

[rphmeier]

not until non-lexical lifetimes are designed and implemented in rust.

[tomusdrw]

Test would be nice :)

[NikVolf]

I believe that any new logic introduced without a test that triggers it should be considered a major issue

[gavofyork]

@tomusdrw please review change in tests/logic carefully.

[coveralls]

Coverage increased (+0.2%) to 87.161% when pulling 1c1fe81 on tx-gas-price-prioritisation into 070a215 on master.

[coveralls]

Coverage decreased (-0.05%) to 86.917% when pulling 7576f4f on tx-gas-price-prioritisation into 070a215 on master.

[tomusdrw]

AFAIR enforce_limit is called everytime transaction is inserted. Wouldn't enforce_limit remove a transaction anyway?

So behaviour in this PR would be: favor transactions that were inserted prior to the limit, when limit is reached you need to provide sufficient gas price to (potentialy) replace one transaction from the queue.

Is this correct/expected?

[tomusdrw]

I would check exact error here if possible:

assert_eq!(res, Err(TransactionError::InsufficientGasPrice {
                minimal: <min>,
                got: <gas_price>,
            }));

[gavofyork]

yes - that's the expected behaviour.

[coveralls]

Coverage decreased (-0.06%) to 86.911% when pulling 0750b06 on tx-gas-price-prioritisation into 070a215 on master.