31 lines
1.7 KiB
Plaintext
31 lines
1.7 KiB
Plaintext
@node chainqueue-tx
|
|
@section Transaction representation
|
|
|
|
Transactions in chainqueue are chain-agnostic. Their representation should only presume a generalized concept of a chain transaction. Interpretation of what a transaction acutally contains or means is left to the client code of the queue to process.
|
|
|
|
In storage each record is divided into two parts: A @emph{mandatory} part, and a @emph{cache} part.
|
|
|
|
|
|
@subsection Mandatory record
|
|
|
|
This consists of the data required to order transactions, and to actually them to the network. Specifically, it contains:
|
|
|
|
@itemize
|
|
@item @strong{hash} of the transaction
|
|
@item the serialized transaction @strong{payload}
|
|
@item the transaction @strong{nonce}
|
|
@item the queue state. @xref{chainqueue-states, queue state}
|
|
@item the @strong{block number} of a confirmed transaction
|
|
@end itemize
|
|
|
|
|
|
@subsection Cache record
|
|
|
|
With exception of the @emph{nonce}, the @emph{cache} part of the record contains deserialized fields of the transaction, all of which can be reconstructed by the client code intepreting the transaction payload in the @emph{mandatory} part.
|
|
|
|
The primary purpose of the @emph{cache} record is performance improvement. By keeping cached records of the expanded and interpreted properties of a transaction, fewer cycles need to be spent later on when needing to access these derived properties.
|
|
|
|
Keeping the cache also more easily enables conditional ordering and execution when querying data sets, e.g. @emph{@guillemetleft{}the first unsent transaction nonce from sender S.@guillemetright{}}
|
|
|
|
Additionally, the cache records curates some additional token semantics, defining in essence a transaction as @emph{@guillemetleft{}sender S sends X amount of token A as Y amount of token B to recipient R@guillemetright{}}.
|