Consider splitting this tool to ./tx and ./submit

[singpolyma]

Because unix. This would allow transactions generated with any tool to be submitted easily with a simple utility, and transactions generated with this tool could still be easily sumbitted with ./tx | ./submit

[donovanhide]

Hi, had thought about this before :-)

It needs a bit of thought though. At which point is the correct Sequence
acquired and If LastLedgerSequence is to be supported, does that occur in
the tx tool or the submit tool?

On 4 August 2014 15:03, Stephen Paul Weber notifications@github.com wrote:

Because unix. This would allow transactions generated with any tool to be
submitted easily with a simple utility, and transactions generated with
this tool could still be easily sumbitted with ./tx | ./submit

—
Reply to this email directly or view it on GitHub
#8.

[singpolyma]

well, sequence is part of the signed transaction, so it needs to be in tx tool. I haven't looked into LastLedgerSequence yet

[donovanhide]

That's kind of the point, is the job of the tx tool to create a signed
binary transaction ready to submit, or is it a way of previewing a JSON
form of the transaction?

The tx tool could possibly be used to create a batch of transactions for
later use. The submit tool could process the JSON input from a file or
stdin, fill in the correct sequences and possibly add a LastLedgerSequence
limit, encode each transaction, submit it and then wait to make sure each
transaction succeeds. If not, report the failed transactions. Kind of like
what Ripple Labs were probably up against with the ComputingForGood payment
code.

It can get complicated quite quickly... Just need to decide what each tool
does and what its use case is. Perhaps we need a batch tool in addition?

On 4 August 2014 15:06, Stephen Paul Weber notifications@github.com wrote:

well, sequence is part of the signed transaction, so it needs to be in tx
tool. I haven't looked into LastLedgerSequence yet

—
Reply to this email directly or view it on GitHub
#8 (comment).

[singpolyma]

Well, in my use case, the submit tool is on a computer that does not have access to the secret key at all, so the prepare tool would need to prepare signed blobs.

[donovanhide]

Maybe we need a pipeline of tx | sign | submit

With sign doing all the clever stuff and being an optional stage?

On 4 August 2014 15:13, Stephen Paul Weber notifications@github.com wrote:

Well, in my use case, the submit tool is on a computer that does not have
access to the secret key at all, so the prepare tool would need to prepare
signed blobs.

—
Reply to this email directly or view it on GitHub
#8 (comment).

[singpolyma]

Hmm, I could support that seperation, yes. That might be ever better. Then, as you say, you can create transactions to be batch-signed and submitted later as well