designforsf / brigade-matchmaker Goto Github PK

To recreate:

1. User selects a tag for skill / interest / learn skill (e.g. HTML)
2. In the top project, user clicks "Contact Project Lead"
3. In the minmaximized messaging form, these tags do not match the tags in the messaging form attachment. In the messaging form, the tags shoe the hierarchy as opposed to the title.

A design needs to be sketched out and prototyped. Requirements for a configuration system:

• configs are accessible: JSON stored in a top-level directory, not dotfiles
• protecting devs from making mistakes: a default JSON is checked in, all others gitignored
• segmented: define configurations targeting each component
• possibly using a standard node module, but not a requirement

Pull the most recent keywords and categories from a spreadsheet maintained by design.

[x] Convert to bootstrap
[x] Create js function to close/retract the LP and display the navigation
[x] Move text and intent toward prototype
[x] Create js function to re-open landing page
[ ] Toggle switch enabling the user to open/close from navigation

The current UI for selected keywords is rudimentary and not based on designs.

Use this recent screen shot of designs being tested:

In this design, the selected keywords are more compact than the current implementation on staging, have coloring, and also do not indicate hierarchical placement.

Tasks: Confirm with Anita about what her role is with the dev of this subcomponent.

New approach enabling maintainers to more easily manage keywords and categories.

• items slide open to reveal detail

Unknown how to represent sub arrays in unix / text output.

Oz and I kicked off this effort a summit or two ago: https://github.com/codeforamerica/civic-tech-taxonomy

The goal was to create a taxonomy that could be easily shared across brigades and brigade tools to

1. Make it easier for brigades and tools to have a quality taxonomy that continuously gets improved
2. Lay the foundation for tools that could work with project/member data across the network

To this end, we settled on maintaining and publishing the taxonomy in sorted YAML files on GitHub, as this would be

• Easy for various tools and brigades to import and load updates from
• Provide a distributable workflow for accepting and reviewing additions/corrections to the taxonomy via pull requests
• Allow brigades and tool makers to fork the taxonomy for experiments in a visible way while maintaining compatibility and a path to merging

To get started, we dumped the wild data set of technology and topic tags that had been accrued in Code for Philly's site over the years with the hopes of then recruiting volunteers to pair that down. That part never got done. We also aimed to get Travis CI setup for quality checking any pull requests to make the workflow for reviewing and accepting additions to the taxonomy as seamless and reliable as possible

Skills and roles taxonomies were also on the horizon. It looks like your research team has produced a quality taxonomy by building one from the ground-up with researchers.

Things to discuss so that we might unite efforts and resume this project together:

• What forms does your taxonomy exist in currently?
• Can we convert it to a sorted yaml format (the structure/attributes currently illustrated were only a first draft and can certainly be iterated upon)?
• Can your team switch your workflow to importing from this repo as the authoritative taxonomy?

On hover, UI should display details more specific details of the project.

When the user selects a tag for skill / interest / learn skill, these tags do not match the tags in the project list. In the project list, the tags shoe the hierarchy as opposed to the title.

Create a function which can be used to render just the project's taxonomy tags and those specified by the user. Requires a project ID and target div ID to be specified. This is for use within the messaging component.

SEE the following URL:

http://localhost:5465/api/user/matches?skills=client-dev/javascript?learnSkills=?interests=

• define a group prefs data structure
• create a json file to be loaded on startup

Proposed data structure:

• step
• isComplete
• startedAt
• updatedAt
• user
  • email
  • password
  • interests (array of keywords)
  • skills (array of keywords)
  • roles (array of keywords)

Basic functionality:

• nodejs service running independently from the web app
• REST API that can be called from the UI
• message format (includes to "project id", from, message, user selection from taxonomy)
• API call leads to a message being queued
• service periodically examines the queue and sends messages to Slack

After the makes a selection and generates matches, the details pane within each project is not appearing.

Proposed data structure:

• groups (array of hashes)
  • name, github project name
• preferences
  • interests (array of keywords)
  • skills (array of keywords)
  • roles (array of keywords)

Or perhaps renamed to "static".

A Backbone View (possibly called SelectorUI) in the selector_ui component that creates a single API to the three SelectorView, and provides the following functions to the Web App and other components that may embed the Selector UI.

• getSelection()
• indicateMatchingStarted()
• indicateMatchingFinished()

How this might be used in the Web App's project-match.js:

    // called when the user clicks on the generate match button
    PM.generateMatch = function () {

    	console.log('ProjectMatch.generateMatch');
			var selectorUI = ProjectMatch.SelectorUI;

    	// UI: matching started
    	selectorUI.indicateMatchingStarted();

      // search with the current user's taxonomy selection
      projView.searchProjects(selectorUI.getSelection());

      // UI: matching finished
      setTimeout(function () {
      	selectorUI.indicateMatchingFinished();
      }, 1000);
      
    }; // END generateMatch

Create a set of instructions that a nodejs developer may use to get brigadehub running locally.

• define a set of test groups
• initially create a set of command line tests
• implement in mocha

Update list to reflect current UI prototype using bootstrap.

Python script to test out the sorting algorithm.

Input:

• user
  • email
  • interests: comma delimited keywords
  • skills: comma delimited keywords
  • role: comma delimited keywords

Database (initially simulated):

• groups, each having:
  • interests: array of { keyword, weight }
  • skills needed: array of { keyword, weight }
  • roles needed: array of { keyword, weight }

Output:

• sorted list of groups

A data format is defined. Involves https calls to GitHub, then storing the data in the local mongodb instance. This would go through mongoose. This should be an API call, and a cronjon can be created that calls this with wget or curl.

When setting skills=javascript in the UI selector, the results do not reflect this selection. Also, viewing the log I can see that it is not coming through in the ajax / XHR request.

GET /api/user/matches?skills=&learnSkills=&interests=& 304 646.620 ms - -