With RCT, we allow 0 size outputs, to try and encourage txes
with two inputs and two outputs. Consolidation would then
have two non zero inputs, one zero output, and one larger
output.
Before the normal selection, we attempt to find either one or two
suitable outputs to use as inputs to the rct tx. The intent is that
most rct txes will have one or two inputs, and we want all to look
the same if possible.
When two outputs are needed, we try to find a pair which are not
related (ie, by being from the same or similar block height).
The "transfer" simplewallet command is renamed to "transfer_original".
"transfer_new" is renamed "transfer", "transfer_rct" is removed,
and the new "transfer" now selects rct or non rct transactions
based on the current block height.
Since these are needed at the same time as the output pubkeys,
this is a whole lot faster, and takes less space. Only outputs
of 0 amount store the commitment. When reading other outputs,
a fake commitment is regenerated on the fly. This avoids having
to rewrite the database to add space for fake commitments for
existing outputs.
This code relies on two things:
- LMDB must support fixed size records per key, rather than
per database (ie, all records on key 0 are the same size, all
records for non 0 keys are same size, but records from key 0
and non 0 keys do have different sizes).
- the commitment must be directly after the rest of the data
in outkey and output_data_t.
The mixRing (output keys and commitments) and II fields (key images)
can be reconstructed from vin data.
This saves some modest amount of space in the tx.
If the blockchain gets reorganized, all outputs spent in the part
of the blockchain that's blown away need to be reset to unspent
(they may end up spent again on the blocks that replace the blocks
that are removed, however).
It may be suboptimal, but it's a pain to have to rebuild everything
when some of this changes.
Also, no clue why there seems to be two different code paths for
serializing a tx...
A new version of genRct takes the mixRing as parameter, instead
of the inPk. inPk are part of the mixRing, and it is cleaner to
pass the mixRing data than to fetch it from the RingCT code.
A new version of decodeRct also returns the mask.
Also, failure to decode throws, so errors are properly detected.
This lets my gcc picks those instead of the generic template
where appropriate (and then fail since std::vector<something>
does not have a serialize method.
