Reinventing the Transaction Script (NDC London 2020)

Reinventing the
Transaction Script
(NDC London 2020)
@ScottWlaschin
fsharpforfunandprofit.com

Are transaction scripts
a Bad Thing?

Are transaction scripts
a Bad Thing?
No!

Part I
What is a "Transaction Script"?

What is a "Transaction Script"?

Question:
How should you organize
your domain logic?

Transaction Script
• Most business applications can be thought of
as a series of transactions.
• A "Transaction Script" organizes this logic as a
single procedure, making calls directly to the
database or through a thin database wrapper.
• Each transaction will have its ownTransaction
Script

The PEAA patterns for domain logic
• Transaction script: each procedure handles a
single request
• Table Module: one instance handles the
business logic for all rows in a database
table/view.
• Domain Model: An object model of the domain
that incorporates both behavior and data.
• Service Layer: a layer of services that
establishes a set of available operations

The PEAA patterns for domain logic
• Transaction script
• Table Module
• Domain Model
• Service Layer
Too simple, too "anemic"
Too db centric
Nice and complex!

The developer's curse
Simple Complex

Transaction scripts have a bad rep
• The reputation
– It's for people who are stuck in the 1990s
– It's for people who use Cobol
– It's only for uncool people
• The truth
– It can be reinvented using functional programming
– Therefore, it can now be cool again!

The pros and cons of
Transaction Scripts

A typical transaction script
get data from DB
do some business logic
write some data to DB
some more business logic
if some condition then
some more DB access
some more DB access
else
some more DB access
Pro: Easy to understand.
One directional.
Pro: Focused!
All code is relevant.
YAGNI for free

get data from DB
some more DB access
some more DB access
else
some more DB access
Con: Hard to modify

get data from DB
some more DB access
some more DB access
else if some other condition then
else
some more DB access
Con: Hard to modify
and evolve.
No graceful path to
richer behavior

get data from DB
some more DB access
some more DB access
else
some more DB access
Con: Hard to reuse.
No rich domain model.
How do you reuse this bit only?

get data from DB
some more DB access
some more DB access
else
some more DB access
Con: Hard to test
Business logic is
mixed up with DB access

get data from DB
some more DB access
some more DB access
else
some more DB access
Con: Hard to test
How can you test
just these bits?

get data from DB
some more DB access
some more DB access
else
some more DB access
Can we fix
these problems?
Yes, we can!

The fix:
Use Functional Programming
& Domain Driven Design
High coolness score

Part II
Intro to Functional Programming
Function

The Tunnel of
Transformation
Function
apple -> banana
A function is a thing which
transforms inputs to outputs

A function is a standalone thing,
not attached to a class
It can be used for inputs and outputs
of other functions

input
A function can be an output
A function is a standalone thing

output
A function can be an input

input output
A function can be a parameter

Core FP principle:
Composition everywhere

Lego Philosophy
1. All pieces are designed to be connected
2. The pieces are reusable in many contexts
3. Connect two pieces together and get
another "piece" that can still be connected

All pieces are designed to be connected

The pieces are reusable in different contexts

Connect two pieces together and
get another "piece" that can still be connected

Make big things from small things in the same way

Function composition
Function 1
apple -> banana
Function 2
banana -> cherry

>>
Function 1
apple -> banana
Function 2
banana -> cherry

New Function
apple -> cherry
Can't tell it was built
from smaller functions!
Where did the banana go?
(abstraction)

Building big things from functions
It's compositions all the way up

Low-level operation
ToUpper
stringstring

Low-level operation
Service
AddressValidator
Validation
Result
Address
Low-level operation Low-level operation

Service
Use-case
UpdateProfileData
ChangeProfile
Result
ChangeProfile
Request
Service Service

Use-case
Web application
Http
Response
Http
Request
Use-case Use-case

Http
Response
Http
Request
Check out my "Power
of Composition" talk
for more details.

Composition everywhere:
Types can be composed too

New types are built from smaller types by:
Composing with “AND”
Composing with “OR”

Example: pairs, tuples, records
FruitSalad = One each of and and
Compose with “AND”
type FruitSalad = {
Apple: AppleVariety
Banana: BananaVariety
Cherry: CherryVariety
}

Snack = or or
Compose with “OR”
type Snack =
| Apple of AppleVariety
| Banana of BananaVariety
| Cherry of CherryVariety

Real world example
of type composition

Example of some requirements:
We accept three forms of payment:
Cash, PayPal, or Card.
For Cash we don't need any extra information
For PayPal we need an email address
For Cards we need a card type and card number

interface IPaymentMethod
{..}
class Cash() : IPaymentMethod
{..}
class PayPal(string emailAddress): IPaymentMethod
{..}
class Card(string cardType, string cardNo) : IPaymentMethod
{..}
In OO design you would probably implement it as an
interface and a set of subclasses, like this:

type EmailAddress = string
type CardNumber = string
In FP you would probably implement by composing
types, like this:

type EmailAddress = ...
type CardNumber = …
type CardType = Visa | Mastercard
type CreditCardInfo = {
CardType : CardType
CardNumber : CardNumber
}

type CardNumber = ...
type CardType = ...
type CreditCardInfo = ...
type PaymentMethod =
| Cash
| PayPal of EmailAddress
| Card of CreditCardInfo

type CardType = ...
| Cash
type PaymentAmount = decimal
type Currency = EUR | USD

type CardType = ...
| Cash
type PaymentAmount = decimal
type Currency = EUR | USD
type Payment = {
Amount : PaymentAmount
Currency: Currency
Method: PaymentMethod }

Applying FP principles to
Transaction Scripts

Introducing "Workflows"
A.k.a. "Transaction", "Story" , "Use-case"
Check out "Event Storming" for
understanding business events

Implementing workflows
A workflow will be
implemented by a function!

Composable
Type
Composable
Type
Inputs and outputs are
defined by composable types

We will compose a workflow
from smaller components
Each component is a
standalone function

Part III
Intro to Domain-Driven Design

The DDD book came out just
after PEAA in 2003
"Focus on the domain and domain
logic rather than technology"
-- Eric Evans

• Agile contribution:
– Rapid feedback during design
• DDD contribution:
– Stakeholders have a shared mental model
– …which is also represented in the code
How can we do design right?

Key DDD principle:
Communicate the design
in the code

Can you really make code
represent the domain?

What some source code looks like

module CardGame =
type Suit = Club | Diamond | Spade | Heart
type Rank = Two |Three | Four | Five | Six | Seven | Eight
| Nine |Ten | Jack | Queen | King
type Card = Suit * Rank
type Hand = Card list
type Deck = Card list
type Player = {Name:string; Hand:Hand}
type Game = {Deck:Deck; Players: Player list}
type Deal = Deck –› (Deck * Card)
type PickupCard = (Hand * Card) –› Hand
Sharedlanguage What DDD source code
*should* look like
(this is F#)

module CardGame =
type Player = {Name:string; Hand:Hand}
type Game = {Deck:Deck; Players: Player list}
* means a pair. Choose one from each type
list type is built in

Deal
(original)
Deck
(remaining)
Deck
(on table)
Card
Modeling a workflow with a function
type Deal = Deck -> (Deck * Card)
Input Output

Pickup Card
(updated)
Hand
(original)
Hand
(on table)
Card
Modeling a workflow with a function
type PickupCard = (Hand * Card) –> Hand
Input Output

module CardGame =
type Player = { Name:string; Hand:Hand }
type Game = { Deck:Deck; Players: Player list }

module CardGame =
Can non-programmers provide
useful feedback?

In the real world
Suit
Rank
Card
Hand
Deck
Player
Deal
In the code
Suit
Rank
Card
Hand
Deck
Player
Deal

In the real world
Suit
Rank
Card
Hand
Deck
Player
Deal
In the code
Suit
Rank
Card
Hand
Deck
Player
Deal
PlayerController
DeckBase
AbstractCardProxyFactoryBean


Key DDD principle:
Communicate the design
in the code


Introducing "bounded contexts"

Grouping functionality in the domain
A bounded
context

Why "Bounded Context"?
• "Context"
– Specialized knowledge and common language
– Information taken out of context is confusing or
unusable
• "Bounded"
– Contexts must be autonomous so they can
evolve independently.
– Boundaries keep you focused! No scope creep!

Bounded context with workflows
A bounded context contains a set of
related workflows

Bounded context with workflows
A bounded context contains a set of
related workflows/functions

Part IV
Reinventing the Transaction Script

• FP contribution:
– Transactions are functions
– Build them from components using composition
• Functional Domain Modeling contribution:
– Use composable types for a rich domain model
– The domain actions are standalone/reusable
– Use autonomous bounded contexts to group and
manage the workflows
Reinventing the Transaction Script

Pros and cons of Transaction Scripts
Pros
• Easy to understand
• Focused: All code is
relevant.
Cons
• Hard to modify/evolve
• Hard to reuse
• No rich domain model
• Hard to test

Claim:
FP-style transaction scripts
are easier to comprehend than
OO domain models.

FP workflow
All arrows go left to right

Example: a web backend
One directional flow,
even with branching

Object Oriented workflow
Arrows go in all directions
We don't design microservices this way!

Real-world OO dependency graph
Multiple cyclic dependencies 

Real-world FP dependency graph
All arrows go left to right 
Same functionality
as the OO example

Claim:
are resistant to bloat

Some OO-style interfaces
interface IRepository {
Insert
InsertAsync
Delete
DeleteAsync
Update
UpdateAsync
CommitChanges
CommitChangesAsync
Query
QueryByKey
QueryWithFilter
QueryWithSpecification
Contains
Count
QuerySummary
QuerySummaryV2
ChangePassword
ChangePasswordV2
DeleteAllRows
LaunchMissiles
LaunchMissilesV2
}
One interface that supports *every*
possible workflow!
Where's the
Interface segregation principle?

FP workflow
Transaction script style functions
only contain what they need.
Every part is relevant.
You get the ISP for free.

Claim:
can be modified with confidence

Modifying a workflow
Replace a component
Static type checking ensures that
sub-components are used correctly
Minimizing the amount of code that I touch

Add features to a workflow
Insert new logic
Minimizing the amount of code that I touch

Add branching to a workflow
Add conditional logic

Parameterizing a workflow for reuse
Swap out parameters for reuse
as long as input and output types match

Claim:
can have a rich domain model

A rich domain model...
These functions are independent, so these
workflows can still evolve independently
...used in functions

Sharing rich domain components
Shared component
Yes, complex domain logic *is*
compatible with transaction scripts

Claim:
are the natural evolution of
Onion/Clean/Hexagonal
Architecture

Traditional layered model
A change to the way that a workflow works
means that you need to touch every layer.

Vertical slices
Each workflow contains all the code it needs to get its job done.
When the requirements change for a workflow, only the code in
that particular vertical slice needs to change.

Vertical slices stretched out
Confusing!


The "onion" architecture
Core domain is pure, and all I/O is at the edges

See "Functional Core/Imperative Shell"

Claim:
are easy to test

Review of Key Testing Concepts
• The SUT (System Under Test) should be a unit of
business value
– Test transactions, not classes
• Tests should apply to the boundaries of a system
not its internals
– Tests should be done at the transaction level
• A "Unit" test means the test is isolated
– That is, it produces no side effects and can be run in
isolation.
– A unit is not a class!

In a functional design, all
I/O is at the edges.
A FP-style transaction script

Database Database
Load data
Process it
Save it

DeterministicNon-deterministic Non-deterministic

I/O in the middle of a workflow
Keep them separate

FP style transactions work
on the front end too!
event-driven!

MVU is a FP style approach
As seen in Elm, Redux, etc

MVU is a FP style approach
Deterministic
function
Deterministic
function
Non-deterministic
actions ("I/O")

PartV
Deployment options for

3 different architectures…
• For monoliths:
– Each bounded context is a separate module with a
well-defined interface
• For service-oriented architecture:
– Each bounded context is a separate container
• For serverless:
– Each individual workflow is deployed separately

In conclusion:
The ReinventedTransaction Script

The ReinventedTransaction Script
• Most business applications can be thought of
as a series of transactions.
• A "Transaction Script" organizes this logic as a
single function, with a deterministic core
and I/O at the edges.
• Each transaction will have its own,
autonomous, evolvableTransaction Script

In conclusion…
Transaction scripts should be more popular
• Business-focused not technology-focused
• Great for agile:
– "Transaction" as the unit of development
• Easy to understand: dataflow is one directional
• Less bloat. You get ISP andYAGNI for free!

In conclusion…
Problems are solved by
• FP composability
• Separation of I/O
New, improved transaction scripts!
• Have a rich domain
• Are easy to modify
• Are easy to test
• Are microservice and serverless friendly!

"Reinventing theTransaction Script"
– Slides and video will be posted at
• fsharpforfunandprofit.com/transactionscript
Related talks
– "The Power of Composition"
• fsharpforfunandprofit.com/composition
– "Domain Modeling Made Functional"
• fsharpforfunandprofit.com/ddd
Thanks!
Twitter: @ScottWlaschin

Reinventing the Transaction Script (NDC London 2020)

Recommandé

Recommandé

Contenu connexe

Tendances

Tendances (20)

Similaire à Reinventing the Transaction Script (NDC London 2020)

Similaire à Reinventing the Transaction Script (NDC London 2020) (20)

Plus de Scott Wlaschin

Plus de Scott Wlaschin (20)

Dernier

Dernier (20)

Reinventing the Transaction Script (NDC London 2020)