jayrbolton

workflow jgi_read_qc
{
	File raw_fastq
	Array[String] sketch_dbs = [ "nt", "refseq", "silva" ] 

	call subsample
	{
		input:
			infile=raw_fastq
	}

jayrbolton

List-of-counters UI challenge for evaluating frameworks

Stage 1

Create a counter component with

• Some text displaying total count
• An "increment" button
• A "decrement" button

Stage 2

jayrbolton

# A task is a docker image, a command, and optional hardware prefs
# A pipeline is a collection of sequential or parallel tasks with a shared volume mount
# Pipelines are submitted to HTCondor

# A small application layer would handle the job submission to HTCondor,
# pulling docker images and running containers with correct mounts and settings

# Define an HTCondor job with a docker image, command, and node requirements or preferences
subsample = Task(
  image='jgi/subsample',

jayrbolton

# Define a docker workflow (independent of KBase)
# - output and input files
# - minimum node requirements for a job
# - whether to exit on any failure or continue on failure
# - pass through environment variables
# - htcondor, etc backend
# - automatically figure out serial and concurrent execution based on task input and output


def subsample():

jayrbolton

Simple doc fetch

{
  coll: "wsprov_object",
  keys: ["1:2:3"],
  user_ids: [],
  page_limit: 10
}

produces aql:

jayrbolton

const Component = require('..')
const h = require('../h')

// A single counter that can be incremented and decremented
const Counter = function (start) {
  return Component('div', {
    count: start,
    add: ({ count }, n) => ({ count: count + n }),
    view: ({ count, add }) => {
      return h('div', [

jayrbolton

Regexes could be a lot more readable with a little bit more verbosity

• Match characters in single quotes
• Use plain keywords for groups of chars (like whitespace)
• Slightly alter some syntax to make more explicit

Match an exact string with single quotes: 'hello world'

Match either strings: 'hello' | 'world' (matches 'hello' or 'world')

jayrbolton

JSON generated by markdown

One downside of JSON is it is not very readable, especially when used as a configuration language.

To fix this, we can embed JSON data in a markdown file using a certain format that is readable and commentable. A script takes the markdown file and converts it into one or more JSON files.

Define the start of a JSON file in your markdown with a header that is a link to a local json file, such as:

## [my_config.json](my_config.json)

jayrbolton

package-name

jayrbolton

Simple spec for small HTTP JSON APIs using JSON Schema

Throughout the spec we use JSON schema documents

Documentation

/docs serves a list of your endpoints.

["products", "orders", "customers"]