Skip to content

Latest commit

 

History

History
293 lines (216 loc) · 7.12 KB

notes.md

File metadata and controls

293 lines (216 loc) · 7.12 KB

Future Development Notes:

  • only support single additions of completions in CLI
    • later we will support multiple additions from file
  • only support single deletion of completions in CLI
  • explore process.cwd() for command line import
  • es6 import and export using babel? experimental features?
  • move extractPrefixes, index, remove, and valid... to helper fn file
  • DBC
  • think about renaming bump
  • profanity filter / ban list?
  • we should trim incoming strings
  • should we support categories?
  • the same completion with different casing has different scores

Possible Additions:

  • allow batch updating of scores
  • allow specifying scores in import
  • is "search" a good name for our command line instruction?
  • unnegate score before returning?

Soulheart docs -http://sethherr.github.io/soulheart/loading_data/

=====================

Client js:

When the value of input changes get request to app to get suggestions display list of suggestions under input

When suggestion is clicked input value should update to suggestion PUT request to app to increment score

When form is submitted PUT request to app to dynamic increment score

Configuration: how many suggestions to show

=====================

API Endpoints:

##get /completions As a app developer, I want to hit this endpoint with a prefix. I expect the status of the response to be 200 I expect the result to be a json of suggestions for this prefix. Request: { prefix: "Go", limit: 5, withScores: true }

Response: [ { completion: "Goku", score: -10000 }, { ... }, ... ]

Request: { prefix: "Mr. M", }

Response: ["Mr. Mime", "Mr. Magoo", "Mr. Monster", "Mr. McDonald", "Mr. Macaroni"]

##put /increment only increments existing completions in dictionary As a app developer, I want to hit this endpoint with a completion I expect the completion score to be incremented and to be given a 200 to verify that happened and an object that includes the completion and the score

Request: "Mr. Mime"

Response: { completion: "Mr. Mime", score: -3001 }

##put /dynamic-increment increments existing completions or adds it if not existing

============================== ##increment discussion Right now want to simply use increment/dynamicIncrement so we don't have to hold state about whether or not the app developer is using a fixed or dynamic dictionary. Their choice of incrementing endpoint will result in one or the other.

###dynamic dictionary desired endpoints (naive, no bucket limit)

  • increment: will increment existing or add new (handle selections/submissions)
  • import : ["oneword"]

###fixed dictionary desired endpoints (always naive, no bucket limit) we don't plan on supporting buckets for the fixed dictionary style because we don't handle inbound new additions from users

  • increment: will only increment existing, not add new ones

###dynamic dictionary desired endpoints (bucket limit)

  • increment: will increment existing and add new
  • this will need logic for kicking out 300th and adding new ones and so on
  • tabling this approach for now

== we have max 300 in bucket mr. mime in bucket but it becomes less popular and get kicked out of m's bucket but it still is in mr. m bucket user types mr. m and selects mr. mime increment mr. mime in every prefix that should belong to it if prefix doesn't include mr. mime, because it was kicked out, we want to re-add it.

##post /completions (basically zadd for all the prefixes) As an app developer, I expect to insert/update my prefix dictionary with the completions and optional scores supplied

Handles initial loading of data, handles mass updates, handles inserting one new completions, handles updating one completion

Request: adding new completions ["Mr. Mime", "Mr. Magoo", "Mr. Monster", "Mr. McDonald", "Mr. Macaroni"]

OR

adding/updating new/existing completions with scores [ { completion: "Goku", score: -10000 }, { ... }, ... ]

OR

adding a new completion ["new completion"]

or

updating the score of an existing completion [ { completion: "existing completion", score: -300 } ]

Response:

ideally: 202 that returns a queue address as part of location header queue address will return a 201 with payload once finished

first stab at it: syncronous 204 (204 instead of 201 because we are not returning the object) no body

##delete /completions Request: body: takes an array of completions ["Mr. Mime", "Mr. Magoo", "Mr. Monster", "Mr. McDonald", "Mr. Macaroni"] "String"

Response: OK 204 did some stuff

##put /score Request: body: takes a string of a completion Response: OK 200, { completion: "string", score: -213 }

The summary

post /completions # insert/update one or more completions delete /completions # deletes one or more completions get /completions # search prefix "ja" # give me completions["jay", "jane", "jam"]

put /score # change score for a completion put /increment # increment score for a completion put /dynamic-increment # increment score for existing or add new comp

Not providing just a string for post/delete completions, just expect the app developer to be able to wrap their string in an array.

Tiffany: "they should just read the documentation"

==

===================== 2017-11-02 Meeting w/ Brandon

##Ways to write our functional test?

insertCompletions(data) search() check if search returns the data

// testing search (functional test)?

-use client.zadd to add some stuff to the test-redis namespace -call search -expect return value to include what we zadded earlier

Functional Tests

functional test implies reliance on other systems, unconnected components

functional test is basically any test that isn't a unit test or doesn't qualify as an integration test

##Making Async Calls Synchronous

 const test = async function() {
    await search().then((resolve) => {
      // expect whatever to happen
    }, (reject) => {})
  };

Don't actually need then when using await keyword.

async is a keyword used as part of function declaration to make a function asynchronous.

=====================

no broken windows test early test often test all the time actually

"premature optimization is the root of all evil"

// client.zadd('he', 1, 'hello'); // client.zincrby('he', 5, 'hello'); // 6 // client.zscore('he', 'hell', (err, reply) => { // console.log(reply) // console.log(err) // });

// import // input: take a collection of strings ("completions") // output: number of items inserted // side effect: index data into redis // needs to process each completion

// for each completion // subproblem: how to get all the prefixes of a string // input: a string // output: array of prefix substrings // loop from 1 through length of string // always slice 0 through i

// adding a completion to redis // input: array of prefix strings // output: none // sideffect: // for each prefix // zadd prefix 0 completion

// input: completion // output: return number of prefixes updated // side effect: increment the score of the completion

// for every prefix of the completion, // we increment by 1 the completion in its bucket