Add a ledger_get RPC command to fetch a ledger from the network.

[rec]

Stole JoelKatz's change, fixed a crashing issue on incomplete data, changed to use new directory structure.

[vinniefalco]

Why isn't it called "get_ledger" ?

[rec]

Now the entire message is one big string constant.

[rec]

It's not called get_ledger because we already have ledger_accept, ledger_cleaner, ledger_closed, ledger_current (and more).

However, I'm wondering if "get" isn't too generic. What is the difference between ledger and ledger_get? Even though I know this, it isn't so obvious.

[rec]

What about ledger_request or ledger_historical?

[vinniefalco]

What's wrong with [verb] [noun] ?

[nbougalis]

Looks good.

[JoelKatz]

It's called 'ledger_get' rather than 'get_ledger' for three reasons:

1. This is consistent with the naming of all other RPC functions. I don't think we want to rename them all and consistency is important in a public API.

2. This has convenience benefits. For example, putting the commands in alphabetical order groups all the commands that operate on ledgers. Matching things like "Ledger*" gets all the files that contain the implementations for commands that operate on ledgers.

3. That's how the English language works. This name has a category and a name for the individual thing in that category. The category is "ledger", and the specifier for the member of that category is "get". The name follows the same logic as "planet Earth". The word "get" is not a verb here but part of the name of the command, so it is being used as a noun. The grammar follows how words are actually being used, not their normal parts of speech.

[vinniefalco]

I don't agree at all.

[rec]

Planet Earth isn't a good example at all - both of those words are nouns!

English is an SVO language, so you say, "I'm going to get the ledger" and
not, "I'm the ledger going to get." as you would in say German.

BUT we have a convention, it's publicly facing, and we're sticking to it.
The big change I made was from "ledger_get" to "ledger_request" - because
even explaining to my fellow Ripple engineers, they got confused as to the
difference between "ledger" (which immediately returns a ledger if you have
it, otherwise does nothing) and ledger_get, which requests a ledger if you
don't have it - thus ledger_request.

On Sun, Jun 8, 2014 at 8:50 PM, Vinnie Falco [email protected]
wrote:

I don't agree at all.

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

[JoelKatz]

There is really no such thing as a word that "is" a noun. A word's part of speech is determined by usage. You might say Google "is" a noun, but if you use it as a verb, "Let me Google that for you", then in that sentence, it's a verb.

You are quite correct that if you were going to command someone to get a ledger, you'd say, "get [the] ledger". But "ledger_get" is not a command, it's the name of a command. So it should follow the rule for English two-word names.

You might command someone to "report [on] Yugoslavia", but the name would be "Yugoslavia report", As in, "Is that the Yugoslavia report?", where "Yugoslavia" is used as an adjective to describe the specific type of report but is really a noun since it's part of the specific name of that specific report.

One problem with "get" is that it's not clear whether it commands the server to get a ledger from the network for itself or whether it just means the command returns information about a ledger. I agree that "ledger_fetch" is probably a better name. While you could still interpret "fetch" as "fetch locally and return", it at least somewhat suggests going through a retrieval process.

[rec]

On Mon, Jun 9, 2014 at 11:55 AM, JoelKatz [email protected] wrote:

There is really no such thing as a word that "is" a noun.

This is not correct - unless your claim is "English grammar doesn't exist".

A word's part of speech is determined by usage.

Yes. And almost no one uses words like "food", "money", "furniture" as
anything other than nouns.

You might say Google "is" a noun, but if you use it as a verb, "Let me
Google that for you", then in that sentence, it's a verb.

Google is used as both a verb and a noun. Most nouns, like "France",
"sofa", "mailbox" are not usable as verbs by your average English speaker.
If your argument is "someone once used sofa as a verb, so it's a verb"
then you're back to the "There is no English grammar" argument.

You are quite correct that if you were going to command someone to get a
ledger, you'd say, "get the ledger". But "ledger_get" is not a command,
it's the name of a command. So it should follow the rule for English
two-word names.

What sort of "name" is this?

You might command someone to "report [on] Yugoslavia", but the name would
be "Yugoslavia report", where "Yugoslavia" is used as an adjective to
describe the specific type of report.

Conveniently, report is another word that happens to be both a verb and a
noun!

One problem with "get" is that it's not clear whether it commands the
server to get a ledger from the network for itself or whether it just means
the command returns information about a ledger. I agree that "ledger_fetch"
is probably a better name. While you could still interpret "fetch" as
"fetch locally and return", it at least somewhat suggests going through a
retrieval process.

Well, we discussed this already on this thread. The reason I didn't find
this convincing is that if you "fetch a report", if you succeed you end up
with the report.

If ledger_request requests a ledger that isn't in memory already, it
neither fails or returns a ledger. Instead, it "requests" that ledger - at
some point in the future that ledger will probably be available from this
rippled (but even that isn't guaranteed - there are a bunch of very early
ledger numbers for which no ledger exists on the network).

"Request" was the only verb :-) I could think of which had this meaning of
"asking for results which might appear in the future or not at all".
Neither "get" nor "fetch" has the idea that the result might show up at
some time later.

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

 /t

http://radio.swirly.com - art music radio 24/7 366/1000

[JoelKatz]

I am not saying that English grammar doesn't exist. I am saying that a word is a noun when, and only when, it's used as a noun. You can argue that any word that can be used as an X is an X, but then every word is pretty much every part of speech. I can say "Hand me the Yugoslavia report", does that mean "Yugoslavia" is an adjective just because it's possible to use it as one?

The proper name of this command is "ledger_get". Proper names with two words in them are both nouns, just like "David Schwartz" or "Interim Memorandum". It doesn't matter how the two words are sometimes, often, or otherwise used. All that matters, as far as English grammar is concerned, is how I'm using them now. "Ledger get" is a two-word proper name of a command, and follows the rules for other two-word proper names.

[vinniefalco]

Finally we agree on something. "ledger_get" is a proper name. We should not use proper names for commands. Instead they should follow the same patterns that we use for naming functions. Proper names cloud conversations because they are a form of trivia. For example, is a "TrustSet" the set of all trust lines? Proper names for commands amount to nothing more than "pet names" whose idiosyncracies must first be learned instead of leveraging the readers understanding of English.

[donovanhide]

http://martinfowler.com/bliki/TwoHardThings.html

:-)

[alexdupre]

http://bikeshed.com/

:-)

[rec]

I agree with the bikeshedding comment particularly. :-D

If we're down to grammar quibbles, shall we call it "done"?

On Mon, Jun 9, 2014 at 1:20 PM, Alex Dupre [email protected] wrote:

http://bikeshed.com/

:-)

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