Bozzy

Quality Assurance Presentation

Brought to you by the team

Erica D.      |      Ian G.      |      Sophie L.P.      |  Ted M.

         KWRAERBG    AO*EUPB            SOEUF                TED

Stenography—What is it?

Written Shorthand → Machine Shorthand

Type Syllables, Words, Phrases instead of letters

200+ words per minute, fast as you can speak, think.

Less stress, better ergonomics

Open Source ♥︎ Steno

• Steno was traditionally very proprietary
• Hardware, software, and education were very expensive
• But, wait! Open Steno Project is founded
  • Plover, first free steno engine 
  • Hardware is open and more affordable
  • Free educational resources (texts, games)
• Our customer Mirabai Knight, founded the Open Steno Project, and our team member is currently the lead developer of Plover

Stenographic Dictionaries

• Stenographers work with multiple dictionaries
  • One large dictionary
    • covers English (~150,000 entries)
    • oft-called "theory" dictionary
  • Several smaller dictionaries
    • Personal dictionaries
    • Job-specific and specialized dictionaries
• Dictionaries have a configurable order
• A dictionary entry consists of machine input to
  emulated keyboard output.

Enter… Bozzy

• First free standalone stenographic dictionary editor
• Named, by our customer, after James Boswell, right hand man of Samuel Johnson, who published the first comprehensive English dictionary.

• Make editing stenographic dictionaries easier (vs manual editing)
• Ease the conversion between Plover's JSON and RTF/CRE
• Allow users of stenography systems to edit their dictionaries for free

Objectives

Sample Use Case: Open Dictionary

Sample Use Case: Convert Dictionary

Sample Use Case: Edit & Save

Architecture

• JVM Desktop App, on Windows and OSX
• Model View Controller architectural pattern
• • UI built using JavaFX and FXML
  • Controllers written using Scala 
  • Data models used to hold dictionaries, user settings, and data in memory, no database
• Use the operating system's native file system     
  • Read and write dictionary files and user settings files

QA Strategy & Process

• High internal standards
  • Unit test all new features
  • Mandatory code reviews
  • Code style
• Highly engaged customer with personal interest
  • Incremental releases
  • Reviews against requirements
• Large, willing-to-test community

Test Plan

• Continuously...         (everyone)
  • unit test every feature
    • low-level
    • black and clear box testing
  • code review
  • "full-regression" suite, verify all core features still work
    • high-level tests
    • black box
• Milestones...              (Ted, Sophie)
  • basic feature complete (interview with customer)
  • first public beta (Obs. session, design review, public feedback)
  • release candidate (Obs. session, public feedback)

  // You are so jelly of our beautiful test cases.
  "The DictionaryEntry class" should "load a simple JSON entry" in {
    var entry = new DictionaryEntry("\"KOPB/SREPBGS\": \"convention\",",
                                    DictionaryFormat.JSON, "name")
    entry.stroke.raw should equal ("KOPB/SREPBGS")
    entry.translation.raw should equal ("convention")
    entry = new DictionaryEntry(
      "\"TRAOEU/SER/TOPS\": \"Triceratops\",",
      DictionaryFormat.JSON,
      "name"
    )
    entry.stroke.raw should equal ("TRAOEU/SER/TOPS")
    entry.translation.raw should equal ("Triceratops")
  }

Test Environment

Travis CI running tests on every push, building on release

SBT, build tool for Scala & Java applications

Test Data

• The dictionaries to test and verify the system:
  • StenEd (RTF)
    • Simple theory dictionary in RTF
  • Magnum (RTF)
    • Complex proprietary dictionary
  • Plover Main (JSON)
    • GPL theory dictionary, default with Plover
• Our tests are based off of data from these dictionaries, as well as custom entries made to test edge cases.

{
"#*E": "{>}{&e}",
"#*EU": "{>}{&i}",
"#*U": "{>}{&u}",
"#-D": "{^ed}",
"#-Z": "00",
"#240": "240",
"#45/TK-PL": "$45",
"#EUD/KWROPL": "idiom",
"#K*": "{>}{&k}",
"#KR*": "{>}{&c}",
"#KW*": "{>}{&q}",
"#KWR*": "{>}{&y}",
"#R*": "{>}{&r}",
"#W*": "{>}{&w}",
"*B": "B",
"*BG": "{^k}",
"*BGS": "action",
"*BS": "action",
"*D": "{^'d}",
"*E": "{>}{&e}",
"*E/KHREUPS": "Eclipse"
}

Core Regression Suite

The aim is to cover all functional requirements

Scripts should be executable manually,

or through automated testing.

Dictionary Management

• Create a blank dictionary ("new")
• Open an existing dictionary
• Define a stroke for "new" dictionary
• Move stroke between dictionaries
• View diff
• Save changes
• Close "new" dictionary
• Reopen "new" dictionary, verify priority
• Reorder "new" and existing dictionary
• Add same-chord stroke to "new" dictionary,
  verify that priority is handled.

• 1.4
• 1.1, 1.2, 1.3
• 5.1
• 5.2
• 5.3
• 5.4
• 2.1
• 3.1, 3.3
• 3.2

Searching (Filtering, Sorting)

• Load two dictionaries with data
  • One RTF/CRE, one JSON
• Filter dictionaries:
  • By Stroke
  • By Definition
  • By Word Count
  • By Dictionary
  • By Stroke, Definition, Dictionary
• Sort results by word count
• Clear filter, sort by translation

• FR 1, 4.1
  •  
• FR 4.2
  •  
  •  
  •  
  •  
  •  
• FR 4.3
• FR 4.3

Backup / Conversion

• Load two dictionaries with data
  • One RTF/CRE, one JSON
• Add two backup directories
• Run backup
• Convert JSON dictionary to RTF
• Verify unconvertables
• Remove JSON dictionary
• Add new RTF conversion
• Convert new RTF back to JSON
• View diff since backup
• Restore from backup

• FR 1
  •  
• 6.2
• 6.1
• 7.2
• 7.3
• 2.1
• FR 1
• 7.1
• 5.3
• 6.3

Merge

• Load two dictionaries with data
  • One RTF/CRE, one JSON
• Merge RTF/CRE and JSON dictionary
• Verify unmergeables

• FR 1
  •  
• 8.1
• 8.2

Build

• Load Plover `main.json`
• "Build" with text that contains words
  not in `main.json`
• Verify diff after build task

• FR 1
• 9.2
   
• 9.3

Conclusion

• Exciting new product type
• Exciting new language for our team
  • Easy to test
• Excited customer
  • Valuable, frequent feedback
• Open Source means that all is up for scrutiny.
• Test Plan is designed to give us more
  information than we need
• Reiteration is key
• Without quality, the software is useless