Add resolve URI interface for local tables, remote barrage subscripti…

[devinrsmith]

…ons, read csv, and parquet tables

[devinrsmith]

import static io.deephaven.db.tables.utils.ResolveTools.resolve

trades = resolve('dh+plain://104.198.75.160/scope/trades_aggregation')
quotes = resolve('dh+plain://34.69.133.75/scope/quotes')

quotes_and_trades = merge(
        quotes.view("Exchange", "Instrument"),
        trades.view("Exchange", "Instrument"))
    .selectDistinct()
    .naturalJoin(quotes, "Exchange,Instrument", "QuoteTimestamp=Timestamp,Bid,Ask")
    .naturalJoin(trades, "Exchange,Instrument")
    .sort("Instrument", "Exchange")

Test script. Will keep the servers up until this merges in.

[devinrsmith]

Follow up issues, #1482, #1483

[niloc132]

i'm not quite following this builder wiring and where this belongs, but host must be validated to never have : or / in it (or ?)

i think this is true for a few others too (port is safe since it is an int, etc)

[devinrsmith]

Yep - the validation actually happens in the check by constructing the actual URI and then comparing it to the input:

    @Check
    final void checkHostPort() {
        // Will cause URI exception if port is invalid too
        if (!host().equals(toUri().getHost())) {
            throw new IllegalArgumentException(String.format("Invalid host '%s'", host()));
        }
    }

[devinrsmith]

Will add more tests to verify.

[niloc132]

i understand the relationship to the strings above, but what does this actually mean? there doesn't seem to be any actual transport in this PR that does work (or not) on that flag?

[niloc132]

Also: if a "remote uri" has type name "RemoteUri", this should probably be "isTls"? or rename remote uri to RemoteURI and toUri to toURI?

[devinrsmith]

io.deephaven.client.impl.ChannelHelper#channelBuilder(DeephavenTarget) does the work

[niloc132]

need to assert that applicationId cannot end with /field, etc, right? possibly never allow slashes in any of these properties

[devinrsmith]

isUriSafe checks that. I'll add a suite of UriHelper tests.

[niloc132]

would it make sense to expose other URI notions like authority and path here as well, and handle the tostring aspects at this level, so as avoid repeating %s://%s/%s at each level and only offer the "how do i make a path" based on the specific context?

it does seem weird to have in some implementations

• a method that returns the constant MY_AUTHORITY
• a String.format(... call with %s for the authority, as a placeholder, and
• as a parameter to that format() call, actually pass the constant MY_AUTHORITY again, rather than call authority() (or, just inline it in the format string)

[devinrsmith]

I can pull out some common code to DeephavenUriBase for toURI IMO, but the Deephaven URIs themself prefer to present themselves in their more strongly typed form; and there isn't a need to dip down into the actual URI components atm. And actually looking, I'm not even using scheme() so I can get rid of that.

[niloc132]

Tentative approval, a few things I'd ask for in descending order of how much I want them:

• Make naming acronyms consistent: TLS vs Id, Uri.
• Documentation so that others don't have to figure out the "why" of this. It doesn't have to be official docs, consider a readme.md in java-client/uri, or a wiki page, or a new discussion started so that this can evolve more easily as use cases are presented.
• Break it out as an extension to be enabled, so as to not depend on it for every setup, avoid making it integral at this point (as a better/simpler/clearer window into things that could be achieved in other ways)
• Super mildest of all, split out the not-URI/Uri stuff like FlightSession.schema, SchemaHelper, session token refresh, table handle/ticket cleanup into its own PR so it is its own commit

[devinrsmith]

I've tried to be a bit more consistent with naming. Added some documentation about URI and server side of things. Hoping to punt on the enabling / configuration of URI resolution until it's exposed to gRPC. Will add info to #1483