Test Your Code the Sane Way (QA and Tech Debt Pt 3)

This series is about building quality tech products specifically geared toward tech startups.

Here we talk about reducing tech debt via a sensible testing process from unit tests to integration tests, e2e tests, and even visual regression testing.

Transcripts (the following transcripts are auto-generated and may have typos or inaccuracies):

so let’s get into the last thing

you can add to this stack to

dramatically improve the quality of your

Tech and the speed at which your tech

company is able to deliver features to

Market

and that is

um code that tests code arcanium

welcome to a production by Dr Miles

Aaron

CEO and co-founder at arcanium ventures

don’t forget to subscribe subscribe

subscribe

so this is testing now and when we look

at testing there are a lot of things you

can do there’s a lot of hot takes on

this I certainly have my own but I’ll

just say at a high level

that you’ve got unit tests which might

be testing a function or or a set of you

know a function that runs some business

logic and a unit test is going to say

you know this function is supposed to

add two numbers and it’s going to throw

some test examples in there

um maybe it’ll even generate you know 50

or 100 of them randomly if it’s a better

test and say do these two numbers add up

to what they’re supposed to based on

this function and the expected output

um and there’s there’s a lot of depth

and and um

there’s a lot of nuance that can go into

writing unit tests but unit tests are

kind of like level one you’re just

covering does this function do what it’s

supposed to do

level two would be integration testing

integration testing is saying I have a

service it’s talking to another service

um if I ping this service am I getting

back what I’m expecting and uh again

there’s a lot of complexity we can

uncover there but in a basic sense

you’re you’re going beyond the unit of

software and you’re talking about

communication across modules or across

Services

um and when you go to across Services

some people might even consider that

end-to-end testing end-to-end testing is

our third level end-to-end testing is

saying look I don’t really care what

happens in between but I want to know

that when I press this button I’m

getting this result back now a lot of

things can happen in between but I’m

basically testing n10 does it go end to

end and do what I want it to do

um so on the testing front you know it

is a big lift if you’re going to walk

into a tech company maybe you’re a new

CTO or director of technology or the CTO

now and you’re thinking oh I got to get

my tech under control things are slow

and if you just say hey we’re going to

go for a 100 test coverage day one one

you’re gonna demoralize all your

engineers you’re gonna piss off your CEO

because you’re going to be way too slow

to get things out

um

and so what my recommendation is is to

do two things you can look at two things

to know where should I have a test and

then you can prioritize those

um so the two areas are

where’s my code getting touched the most

so this is easy you can go into your

GitHub or your bitbucket and just look

at where what files are getting

committed the most the most likely

things to break are the things that

people are touching so it’s easy don’t

don’t write tests for something that’s

been working for 10 years it works right

but if it’s getting changed every week

you probably want to test in place

um and the second thing you can do is

say what are the most critical things

for my business what is uh what are the

components on the page what are the

pages that are required to make a sale

or to

um it to retain a customer or to get a

customer to refer someone these are the

things that are going to drive your

business

and so Focus first on those highly

critical elements and make sure that

they’re tested whether it’s unit tests

on the business logic integration tests

for API calls or end-to-end tests

um and there’s a there’s another side of

testing that I can recommend for these

early companies to look at although it

might depending where you are in the

journey it might be premature but that

is um the side of visual regression

testing and there’s a a really cool new

suite of tools that are coming out that

allow you to do this in a no code way so

you know if you’re on a budget or your

early stage I’d recommend looking at no

code tools for

um for regression testing or visual

regression testing but the idea here

is that basically you have something in

your deployment pipeline that takes an

image of every page when it comes out

you can tell it what where to look and

it says

um

upon every new piece of code that comes

out let’s check let’s take a new

snapshot and let’s compare those two

snapshots and see what does and doesn’t

line up and let’s flag that now you can

have this prevent a pull request from

going through so you have complete

control over the quality there or you

could just have it flag it and you can

decide do I wanna is that does it matter

to me that this changed I mean look

you’re going to be changing your UI so

of course changes are going to happen

you have to let some changes through but

it just brings to the surface like oh

you know these two things were coupled

and I didn’t realize and so I changed

the button over here and QA passed it

and it went into production and my users

are seeing it but I forgot that button

is getting used on this page and the

padding change now it broke a layout now

it looks terrible this happens all the

time visual regression testing is a

low-cost easy way

of at least getting awareness of where

those things might be breaking and the

no code tools that are coming out go a

step further many of them and

get into this kind of automated testing

sphere

um you know previously this was done

with like Puppeteer and selenium but um

there are tools that that make it much

cheaper to get into where you’re

basically going to say hey there are

these user actions I’ve got an ID on a

button user is going to scroll down x

amount they’re going to click that

button then this thing is supposed to

happen and you kind of get a sequence of

events that you can program to tell your

testing software hey when I release this

I want to make sure these 50 critical

user actions run perfectly no matter

what code we’re putting through no

matter what new feature we’re adding so

that’s another way that you can clamp

down on the quality