Fit Specification,
v1.1
Fit is a tool for improving communication
between customers, programmers, and
testers. It allows the customer to describe
his or her requirements in a way that can be checked
against the actual program.
The technique Fit uses is called
"specification
by example." It allows customers to
create a document
with tables that describe specific
examples of how the software should work. Fit looks at the document
and checks to
see if the software works the way the tables say it should.
Fit uses small programs, called
"fixtures," to connect
tables to software. Programmers on the development team are responsible
for creating those fixtures. Customers
are responsible for creating the tables.
Testers work with both groups of people to help make sure that all situations
are adequately covered. The three groups
work together, interactively, to create
great software.
This document, and the documents
it references, describes
Fit using specification
by example. It is, itself, an example of how to use Fit. If all the pass cells
in the table below are colored green, then you
can be confident
that your copy of Fit meets this specification.
Using Fit to test itself, as we do here, leads to circular
logic that can
be unraveled more easily by the computer than
by humans. To avoid a headache,
ignore the fact that this document
is written in HTML or that the tables use Fit fixtures. Instead, just think of it as a regular document--one
that happens to be colored green and red.
Functional changes
to Fit that are new in this version of the spec
are highlighted yellow. Layout, formatting, and other non-functional
changes to the specification
are not highlighted.
Details
Fit knows how to do these things:
|
fat.ReferenceFixture
|
|
|
|
|
Description
|
Location
|
Result()
|
Note
|
|
Parse HTML documents
|
parse.html
|
pass
|
|
|
Modify HTML documents
|
annotation.html
|
pass
|
|
|
Interpret tables using fixtures
|
fixtures.html
|
|
incomplete
|
|
Provide a command-line interface
|
cli.html
|
|
incomplete
|
Click the links for details about how Fit
does each of these things. This specification
is a work in progress and the fixtures and command-line
interface portion are incomplete. Implementers: you may ignore these sections
for now.
Fit is available for many programming languages. The sections
above apply to all implementations of Fit.
Extensions specific
to this implementation of Fit are described
here:
|
fat.ReferenceFixture
|
|
|
|
Description
|
Location
|
Result()
|
|
Implementation-specific
extensions
|
extensions.html
|
pass
|
Known errors and omissions, to be fixed in a future version:
- The
<wiki> tag is not specified.
- Parsing-related
errors are described but the errors
arent specified.
- Nested
tables are not specified
- Argument
list (next to fixture name) is not specified.
- Parser
can handle any XML heirarchy, but
only <table>, <tr>,
<td> is specified.
- Files
are expected to be CP1252, but this is
not specified.
To look at in a future release:
·
Why is info text coming
out with a line break when this spec is
run? (Because
Word creates empty cells
with this markup: <p> </p>. Accomodate this?)
- Possible
bug: when a variable column has a blank cell
(in a ColumnFixture), the value is filled in but
it isnt shown in grey.
- More
flexibility in charset handling.
Possible feature changes:
- Add
blank keyword?
- make fat.ReferenceFixture
a standard fit fixture?
- Provide
a way for ColumnFixture to easily annotate a cell
(including when it passes)
- Data
setup fixture, similar to RowFixture[?]
Change History
Changes in version 1.1:
·
HTML to text: Added support for class
attribute in MS Word <p> tag parsing
·
HTML to text: Clarified handling of
leading and trailing line breaks
·
Text to HTML: Changed handling of backwards
line feed
·
Fixture loading: Added specific
error messages when fixtures dont load
·
Standard annotations: Changed all HTML to
put quotes around color attributes
·
Standard annotations: Changed error and
info annotations to convert new text to
HTML.
·
Standard annotations: Changed wrong and
error annotations to strip mark-up from original cell.
·
Standard annotations: Changed error annotation to use standard font size
for error message.
·
Malformed HTML Parsing: Changed unclosed table, tr, and td tags to generate an error.
·
ColumnFixture:
Changed headers to allow multiple words. (not yet specified.)
·
Fixture
loading: Added argument list by fixture name. (not yet specified.)
·
Parsing:
Added support for nested tables.
(not yet specified.)
Changes in version 1.0:
·
Rewrite of the spec.
·
HTML to text: Changed entity to combine
with other leading and trailing whitespace.
·
HTML to text: Changed non-breaking space
character to not
combine with other internal whitespace.
·
HTML to text: Added conversion
of <br> and <p> tags to line-feed characters.
·
HTML to text: Added conversion
of smart quotes to regular quotes.
·
Text to HTML: Added conversion
of line-feed characters
to <br> tags.
·
Text to HTML: Added conversion
of adjoining spaces to entities.
Initial version:
·
Experimental specification
created by Ward Cunningham
and Jim Shore.