"We Ran Out Of Columns" - The Worst Codebase Ever

Summary

The video script narrates the speaker's experience with a chaotic yet fascinating codebase that started with a database column limitation, leading to creative solutions and humorous anecdotes. It touches on the culture shaped by databases, the quirks of legacy systems, and the ingenuity of developers who carved out their own efficiencies within a monolithic application. The story highlights the beauty in the mess, celebrating the direct impact developers had on improving user experiences, despite the lack of a coherent master plan.

Takeaways

• 😲 The speaker expresses excitement over a story that begins with a database running out of columns, highlighting the unusual and challenging nature of the situation.
• 🔢 The script discusses various database limitations, such as a hard limit of 32,000 columns per table in SQL Server and the concept of 'dynamic pivot' when needing more than 1,124 columns.
• 💡 The speaker reminisces about their first software job, which was both a trial by fire and a memorable experience, despite the codebase being the worst and best they have ever encountered.
• 🏢 The importance of the database in a company's culture is emphasized, with the database being described as a 'culture maker' and a central point where all code meets to operate the system.
• 📈 The story illustrates the evolution of database design, from a table with 1,500 columns to using JSON storage in a column, reflecting the creative solutions developers come up with when faced with limitations.
• 🚀 The speaker shares anecdotes about unique and sometimes absurd database practices, such as a table named 'sequence key' used for generating IDs before auto-incrementing was supported.
• 📅 A humorous situation is described where a company's login system was dependent on a manually filled-out 'calendar' table, which if not updated, would prevent logins.
• 👷‍♂️ The narrative includes tales of interns and developers creating one-off applications and workarounds, sometimes resulting in more complex issues or workarounds themselves.
• 🛠️ The speaker reflects on the chaotic yet innovative nature of the codebase they worked with, which lacked a master plan but allowed for individual developers to create their own solutions.
• 🌐 The script touches on the idea of open-sourcing old software for AI training, suggesting that real production code could be beneficial for AI learning.
• 🛑 The story concludes with the speaker's appreciation for the 'beautiful mess' of the codebase, highlighting the positive aspects of working in a system without strict architectural constraints.

Q & A

What was the initial excitement about the story involving columns?

The excitement stemmed from the unexpected situation of running out of columns in the 'Merchants' table, which led to the creation of 'Merchants 2', sparking interest in the challenges and solutions related to database schema design.

What is the maximum column count allowed in SQL Server tables?

SQL Server has a hard limit of 1,024 columns per table, which is a significant constraint that developers need to consider when designing database schemas.

Why is it recommended to reconsider your approach if needing more than 1,124 columns in a dynamic pivot?

Needing more than 1,124 columns suggests that the data model might be overly complex or not normalized properly, indicating a potential need for a more efficient data structure or redesign.

What was the speaker's realization about professional programming compared to their early hobbyist programming?

The speaker realized that professional programming can involve working with poorly designed codebases and dealing with complex, real-world problems, which is quite different from the simpler, more controlled environment of hobbyist programming.

What was the cultural impact of the database in the speaker's first job?

The database became a 'culture maker', influencing how the entire company operated, with everything revolving around stored procedures and the database's constraints, shaping the company's workflow and practices.

What is the significance of the 'sequence key' table in the script?

The 'sequence key' table, with a single column and row, served as a simple yet effective method for generating unique IDs for new entities in the system, highlighting a clever workaround for the lack of auto-incrementing IDs in SQL Server.

Why did the speaker suggest that every piece of software over 10 years old should be open-sourced for AI training?

The speaker believes that real production code from older software could provide valuable insights and learning opportunities for AI systems, potentially improving their ability to understand and work with complex, real-world codebases.

What was the 'calendar' table's role in the system mentioned in the script?

The 'calendar' table was a manually filled-out database table that, when full, would prevent users from logging into the system, illustrating an unusual and potentially problematic dependency on a simple data table for system access control.

How did the speaker describe the experience of working in a codebase with no master plan?

The speaker found it liberating to work in a codebase without a master plan, allowing developers to create their own 'world of sanity' within the larger, chaotic system, and build new solutions without the burden of adhering to a rigid design.

What was Justin's approach to improving the merchant search page's performance?

Justin decoupled the page by making each box on the screen its own endpoint, loading everything asynchronously, which drastically reduced the page load time from minutes to seconds, demonstrating the power of a focused, independent approach to optimization.

Outlines

00:00

😅 Database Column Limit Excitement

The speaker expresses excitement over a story that begins with a database running out of columns. They discuss the maximum column limits in various database systems, such as Microsoft SQL Server, and humorously suggest that needing more than a certain number of columns is a sign to rethink one's approach. The narrative reflects on the speaker's early programming experiences and the stark contrast between amateur and professional codebases, highlighting the chaotic yet fascinating nature of legacy systems.

05:01

📚 Legacy Database Culture and ID Generation

This paragraph delves into the cultural impact of databases within a company, using the term 'Watering Hole' as a metaphor for where all code converges. The speaker shares anecdotes about a 'Merchants' table with over 1,500 columns and the creative workarounds used when the database ran out of columns, such as storing JSON in a column. They also touch on the use of sequence keys for ID generation in the absence of auto-incrementing features in SQL Server, and the resulting implicit joins across multiple tables.

10:04

😵‍💫 Bizarre Database Practices and Intern Stories

The speaker recounts peculiar database practices, such as manually filling out a 'calendar' table that dictates system accessibility and the daily dropping and repopulating of an 'employees' table. They express concern over the lack of automation in these processes and share a story about an intern who had to manually extend the calendar's range to prevent system lockouts. The narrative also touches on the idea of open-sourcing old software for AI training purposes.

15:05

🤯 Over-normalized Databases and Developer Autonomy

The paragraph discusses the concept of 'normalization syndrome' where databases have an excessive number of columns, leading to overly complex and atomic data structures. The speaker shares personal experiences with such databases and the tendency to create one-off applications for individual users, as exemplified by a developer named 'Gilfoil.' The narrative highlights the autonomy developers had in carving out their own areas of sanity within the monolithic app, leading to a decentralized yet functional development environment.

20:06

🛠️ Codebase Chaos and Creative Problem Solving

The speaker reminisces about a particularly chaotic codebase that lacked a master plan or architectural review, allowing developers to create their own mini-apps within the larger system. They describe the ingenuity of a developer named Justin who dramatically improved the performance of a slow merchant search page by refactoring it to load asynchronously. The narrative celebrates the beauty of a codebase that, despite its messiness, enabled developers to solve problems creatively and efficiently.

🌟 Embracing the Beauty in Chaotic Codebases

In the final paragraph, the speaker reflects on the positive aspects of working in a seemingly disorganized codebase, appreciating the lack of constraints and the direct connection with users. They express a sense of nostalgia for the fast feedback loops and the ability to build and iterate quickly without the burden of grand planning or consistency requirements. The speaker finds beauty in the autonomy and creativity that such an environment fosters, even in the face of what could be considered a 'worst-case' scenario in software development.

Keywords

💡Codebase

A codebase refers to the entire collection of source code for a software project. In the video, the codebase is described as both the 'best' and 'worst' the speaker has worked with, highlighting the complexity and legacy nature of the system, which is a central theme of the video.

💡Database Normalization

Database normalization is the process of organizing data to minimize redundancy and dependency. The script mentions 'properly normalized tables with all the foreign keys to Merchants,' illustrating the contrast between a well-structured database and the chaotic, large tables that the speaker found in their first job.

💡Stored Procedures

Stored procedures are a type of database object stored in the database that contains a series of SQL statements to perform a specific task. The script refers to a company culture built around stored procedures, indicating a reliance on this technology and its impact on the company's operations.

💡Dynamic Pivot

A dynamic pivot in the context of databases refers to a technique used to rotate rows into columns or vice versa. The video mentions a limit of 1,124 columns in a dynamic pivot, suggesting a point at which the complexity of the database operations becomes unwieldy.

💡Merchants Table

In the script, the 'Merchants' table is a central database table that stores customer information and has run out of columns, leading to the creation of 'Merchants too.' This highlights the challenges of managing large amounts of data in a single table and the creative solutions developers come up with.

💡JSON

JSON (JavaScript Object Notation) is a lightweight data-interchange format that is easy for humans to read and write and for machines to parse and generate. The script mentions storing JSON in a database column as a workaround, which is a common practice but also a point of contention in the video.

💡Sequence Key

A sequence key in databases is typically used for generating unique IDs. The video describes a 'sequence key' table with a single column and row, which was used to generate IDs before auto-incrementing features were common, indicating a historical approach to database management.

💡Soap Services

SOAP (Simple Object Access Protocol) services are web services that use XML for messaging. The script mentions a dozen SOAP services alongside the main codebase, indicating the complexity and the integration of different technologies within the system.

💡Reflection

In programming, reflection is the ability of a program to examine or 'introspect' its own structure, particularly through the size, type, and properties of an object. The video describes using reflection to dynamically build a string structure based on the database, showcasing a creative but complex solution.

💡Justin's Improvement

Justin's Improvement refers to a specific instance in the video where a senior developer named Justin improved the performance of the merchant search page by decoupling elements and loading them asynchronously. This example illustrates the power of individual initiative in tackling systemic issues within a codebase.

💡Enterprise Design Pattern

Enterprise design patterns are reusable solutions to common software design problems that occur in large, complex, enterprise-level applications. The script contrasts the speaker's experiences with the 'Enterprise design pattern' and the more ad-hoc, direct approach taken in their first job, highlighting a preference for simplicity and direct feedback loops.

Highlights

Excitement over a database issue where they ran out of columns in the 'Merchants' table, leading to the creation of 'Merchants 2'.

Discussion on the maximum column count allowed in SQL Server, with a hard limit of 32,000 columns per table.

Advice to reconsider design choices if needing more than 1,224 columns in a dynamic pivot.

Reflections on the early days of programming and the evolution of professional software development.

Critique of the myth that modern startups use the greatest technologies and have well-structured codebases.

Experience from the first software job and the contrast between personal projects and professional codebases.

The 'Merchants' table's significance in the system, acting as a central data point with over 500 columns.

Innovative workarounds for database limitations, such as storing JSON in a column.

The 'sequence key' table, a minimalist design with a single key and value for generating IDs.

Challenges with a 'calendar' table that manually controlled system access, highlighting the importance of database design.

Humorous account of an intern's task to pre-fill the 'calendar' table for five years to prevent access issues.

Description of a legacy system where the database's design influenced the company's culture and operations.

The 'MacBook Under the Stairs' analogy, illustrating the neglect and potential risks of outdated or poorly managed systems.

The concept of 'database as a culture maker', emphasizing the central role of databases in shaping business processes.

Anecdote about a salesperson gaming the system by manipulating sales records, leading to accounting irregularities.

The story of 'Gilfoil', a legendary developer known for not checking in code and creating one-off applications.

Reflection on the beauty and chaos of a codebase without a master plan, allowing for creative problem-solving.

Justin's improvement of the merchant search page by breaking it into multiple end-point loads, significantly increasing performance.

Nostalgic appreciation for the direct connection with users and the simplicity of problem-solving in the early days of software development.

Transcripts

I have never been more excited about a

story that starts off with we ran out of

columns this is just making me excited

wow flashbang flashbang all right we ran

out of columns the best worst code base

on the merchants 2 table yeah we ran out

of columns on Merchants so we made

Merchants

too oh my gosh what is the allowed a

maximum column uh column count uh squeal

light how about squeal light isn't it

999 oh there's a hard limit of 32,000

okay columns per table 1,600 oh yeah

squeal server you're right you're right

uh Microsoft SQL Server I forgot about

this one okay they have the same thing

if you need more than this in your

Dynamic pivot time to get a second

thought to what you're doing I love this

if you need more than 1,24 columns in

your Dynamic pivot it's time to give a

second thought of what you're doing

really you know I would probably put

that number way lower okay we're talking

about way way low lower than that it's

that 2 to the 10 that things got out of

control you're like really two to the 10

that's the number that's crazy when I

started programming as a kid I didn't

know people were paid to program even as

I graduated high school I assumed that

the world of professional development

looked quite different from the code I

wrote in my spare time it's true it

looks worse that's the big lie I think

our industry has had for so long is that

you take these companies and whether

it's it used to be the fangs of old but

it's now probably the startups of today

uh all these startups using all the

greatest Technologies all the hooking up

to where everybody's just a cyber Mario

that just simply puts pipes together and

that's it even their codebases with

almost no business logic are absolute

dog water when I lucked my way into my

first software job I quickly learned

just how wrong and how right I had been

my first job was a trial by fire to this

day that code base Remains the worst and

the best codebase I've ever had the

pleasure of working in while the

codebase will forever remain locked by

proprietary walls of that particular

company I hope I can share some of its

most fun and scary stories I'm excited

here okay we're excited here come on the

database lives forever in a large Legacy

system the database is more than a place

to store data it is a culture maker it

dude this is so true oh my gosh I never

even thought about that but in my head I

was thinking about my first job my first

real job my first fully real real job in

which everything was stored procedures

and like we we we the entire company

like morphed itself around this idea of

stored procedures it became the the

culture of the company the database sets

the constraints for how the system as a

whole operates it is the point where all

code meets the the database is a water

is the watering hole in our case The

Watering Hole had quite a bit of

pollution it always does man did you

know that squeal server has a limit on

the number of columns you can have in a

table me neither at the time it was 10

uh two to the 10 uh today it appears to

be 2 to the 12 needless to say most

people don't know this we did the reason

Merchants our table store customer

information ran out of columns a long

time ago you know I've seen people get

around some of this by just storing Json

you know how many times you've seen Json

just stored in a random column how is

that legal Jason called it dude

it's oh my gosh something about that

just makes me emotionally painful then

you know when it gets real bad when you

see people using those like those fuzzy

those fuzzy matches or redx matches on a

wear statement to grab to like just look

at the

string that is Json not not we're not

talking about postgres with the uh the

Json good stuff we're talking about not

postgres and and just getting it you

know Jason B yeah it's called No squeal

Jason B is legit though yeah I haven't I

haven't tried any of those yeah columns

like wears Food Dude yes yes this this

is real this is real we column food dude

I've seen this and this is awful it's

2024 and if you're not using postgres

you're either a bank or wrong well there

is squeal light can we all agree squeal

light is amazing I've never said

anything bad about postgres but squeal

light is also very very good hasht ad

squeal light is is very very good squeal

light is slow when people say that I

don't think they know what they're

saying there's an entire talk about how

how not slow squeal light actually is

sharper guy do you write JavaScript on

the server lot of people people be

complaining about speed yet I don't see

them using C or Zig to write the world's

most optimized code that's what I'm just

say is you know you you can't be

complaining about the difference between

like two microseconds when you're out

there using a garbage collector okay

buddy okay

buddy I like serverless it's like okay

oh I'm really concerned about

performance then

anyways needless to say most people

don't need to know this we did the

reasons Merchant 2 blah blah blah

Merchant merch mer 2 was the solution a

table if I remember correctly with 500

plus columns itself damn damn they had

1,500 columns merchants and my best

friend Merchants too were the lifeblood

of the system everything made its way

back to Merchants somehow or other but

it wasn't as if Merchants was the solo

or Duo table there was plenty of

properly normalized table with all the

foreign keys to Merchants but one will

always hold a special place in my heart

sequence

key I can't wait for this sequence key

that for ease of understanding I have

recreated the whole sequence key table

above yes you read that

correctly this is the whole table a

table with the single key and a single

value if Simplicity is a virtue then one

might declare sequence key to be the

perfect table what could be simpler a

constant one in which you don't have to

look up you may be asking yourself what

possible use case could you have for a

table with one column in a row

generating IDs now the story I heard at

the time was the Once Upon a Time squeal

server didn't support Auto incrementing

IDs it was their primary key gener like

I honestly my brain couldn't even

conceive of this many DBS did like this

is just too painful sequence key I just

didn't want to believe it there's no way

I was going to let myself believe that

this was real I just I even as I read it

I'm like n nope not going to do it it's

not gonna be primary key not gonna be

nope

this was the accepted correct

answer my search to figure out if this

was true was inconclusive but in

practice had served as much as much more

than that sequence key was the glue in

every stored procedure that created new

entities you first grabbed a key from

sequence key incremented it then insert

that as your ID for n different tables

you now had an implicit join between all

those entry tables if you saw an ID in

the system there was a good chance

related tables would have a row with the

exact same ID honestly kind of clever

but isn't that just like database the

calendar a database may live forever but

our login system was limited by the

calendar I this okay okay I don't mean

an actual calendar I mean a database

table called calendar what did it

contain a manually filled out calendar

when asking a resident Shaman who went

by the name of munch he informed me that

when the calendar runs out we can't log

to uh log into the system I got to know

why that is I got I just got to know I

just got to know how that happened like

how how did that happen this happened a

few years ago so they had an intern fill

out five more years to make sure that it

didn't happen anytime soon what system

used this calendar no one knew what also

like couldn't the intern just write a

for Loop of inserts like do you even

have to man like now I'm concerned oh

now you're concerned so let let me get

this straight you weren't concerned when

the things called we ran out of columns

you were

concerned when they had an intern that

that that filled in five more years for

logging in MacBook Under the Stairs

again this is MacBook Under the Stairs

except for Worse what's a for Loop you

can for days nuts

uh employees every morning at 7:15 the

employees table was

dropped you think layoffs are bad now

you should saw layoffs back then okay

this was this is the real one all the

data completely gone then a CSV from an

ADP was uploaded into the table during

this time you couldn't log into the

system sometimes this process failed

this wasn't the end of the process the

data needed to be replicated to

headquarters so an email was sent to a

man who every day would push a button to

copy the data one would just think you'd

have

you would one would think this process

only happened when there was a change in

employment you know what you know what I

think would be really great I think

every company you know what hey since

we're since we're we're talking about

video games stop killing video games and

all this you know what I think I think

we need to have every single piece of

software that's 10 years older or more

to be open sourced and for the llms to

train on okay I want real production

code for all these llms to be trained on

I think that's the only reasonable way

to get things done because I don't see

how I if co-pilot does not suggest this

to me I'm going to be really really

upset I think co-pilot should suggest

the email once a day press the button to

deploy the employees

table every day okay and if this guy

goes on vacation all hell's breaking

loose okay that's what I want the

replacement database you might be

thinking to yourself couldn't someone

clean up this database make it nice to

nicer to work with dude no did not think

that to myself cuz there's 1,24 columns

in merchants and over 500 in Merchants

too that's an imposs no one's cleaning

up that code base ever for any reason

ever forever ever ever ever okay well

the company was way ahead of you there

was a copy of the database data in this

copy was about 10 minutes out of date

syn only went one way but this database

was normalized how normalized to go from

Merchants to a phone number was seven

joins Somebody went real Enterprise you

know I you know the funny part is is I

know everyone here is like damn how

could they do that oh this is awful but

I've done that you know there's this

thing that happens where you first work

in an environment in which you have

exceeding amount of columns and so the

answer is to make everything Atomic

normalization syndrome yeah this happens

this isn't somehow wild right why is

this so relatable to me because this is

everybody's mistake with databases and

you know what if I were to start doing a

bunch of database work again guess how

many of these mistakes I would

make I probably wouldn't get out to the

maximum number of columns but I could

easily see myself going I'm at like

25 columns at this point maybe I maybe I

need to rethink my maybe I need to

rethink my stuff and then it keep go up

to like 35 I'm like damn when you first

move into a house it's shiny and new and

then you do things and then you start

doing things to it the sales number

every salesperson had a quota they

needed the meet every month called win

oh is this Cloud wait is this a cloud

flare like is this the are we about to

meet the saf the sa the safety and Trust

team commonly referred to as the sales

team the tables that kept this data not

the financial keepings but the sales

specific way of accounting for this were

incredibly complicated every day a job

would figure out which rows had been

added and updated and sync them with the

system at headquarters if this wasn't

really a problem until one salesperson

figured out they could ask for those

records to be manually changed this

salesperson had already got their win

and landed another big sale that month

so they wanted it to be moved to next

month oh my gosh what a genius the man

is stacking sales oh this is so good

imagine accounting for this like just

just for a second this guy just effed up

Accounting in a whole special beautiful

way like so so beautiful you're

receiving money but no sale has happened

for a while oh oh this is so good this

is so this is this is the perfect storm

of of gaming the system and inter inter

was tasked with doing so word got out

and over the next 3 years requests would

grow exponentially at one point we had

three interns whose fulltime job was

writing these squeal statements writing

an application to do this was deemed too

difficult before I left though I made

sure to help those interns build their

own no idea if it ever took off though

IRS shambles deeply illegal accounting

actually I know this is deeply illegal

this is so good the code base what but

what is a codebase without or a database

without a code base and what a

magnificent codebase it was when I

joined everything was in team Foundation

server oh I I've heard nothing but good

things about team Foundation server you

know I've never had the privilege of

working in it myself but I've heard

amazing stories that it's absolutely one

of the best pieces of software out there

ever if you aren't familiar this was

Microsoft made centralized Source

control system the main code base I

worked in was half VB half C now that

seems pretty normal I actually I I've

worked in very similar ones it ran on

IIs oh man I now that's actually the

only server that I was not smart enough

to understand sure I could raw dog TCP

created my own video format was able to

send uh I had a whole distributed graph

Network to be able to send uh my own

custom protocol through nodes that the

for the government uh Hardware hopping

and all that couldn't understand I I I

it it literally was too it was too

confusing couldn't do it literally could

not do it we still use IIs bless your

soul and use session state for

everything oh this this actually seems

like a good idea I would use session

state for everything engine x a riot oh

what did this mean in practice if you

navigated to a page via P path a or path

B you'd see very different things on

that page but to describe this codebase

as merely half VB and half C would do a

disservice every JavaScript framework

that existed at the time was checked

into the repository typically with some

custom changes the author believed

needed to be made most notably knockout

backbone and marionette but of course

there was a smattering of jQuery and

jQuery plugins m nothing like knockout

backbone marionette Jacqui and some

jQuery plugins how much do you want to

bet date Pickers in there date picker by

the CEO of frontend Masters Mark good

guy Mark making date picker for

everybody's life date picker was the

single greatest library of all time okay

don't don't even try to tell me that

jquery's date picker wasn't

revolutionary don't even try don't even

try but this codebase didn't stand alone

next to it were a dozen or so soap

services and a hand full of native

Windows applications most notable was

the shipping manager Fable has an entire

application was built in a weekend to

buy a solo developer let's call him

gilfoil gilfoil was by all accounts an

incredibly fast programmer I never met

him but I felt like I knew him not just

through his code in the repos but also

through all the code remaining on his

hard

drives wait what Gil foil's hard drives

wait what Munch yes this was the name he

really went by kept Gil foil's hard

drives in in raid configuration on his

desk years after Gil foil had left the

company why because Gil foil was known

for not checking in code not only that

but for building a random one-off

Windows application for a single user it

wasn't uncommon to have a user come to

us with a bug report for an application

that only existed on Gil foil's hard

drive I can NE I cannot say that this

I've have ever experienced anything like

this this is this is a peak of

coding and athleticism I have I have

never seen okay it's too good it's

actually too good for me he's literally

the goat okay bus Factor

one deny refuse literally customers

coming and be like hey I need to fix

this bug on this application what

application we've never heard of that

well actually it turns out one of you

guys wrote an application just for me

most of my job was tracking down bugs

that teams didn't want to dedicate work

to one particularly nasty bug would pop

up every once uh once every few months

after we shipped things the shipping

queue would have would have stuck orders

in them that claimed to both be already

shipped and not shipped I went through a

series of workarounds a squeal script

window applications Etc to try and get

us out of the broken State I was advised

not to try and track down the root cause

but I couldn't help myself oh yeah oh

yeah along the way I learned how Gil

foil thought the shipping app pulled

down the entire code base or the entire

database and then filtered by date

keeping all past orders the go uh oh

keeping all orders past the go live date

of the application the app relied on a

soap service not to do any servy things

no the service was pure function it was

the client that did all the side effects

in that client I discovered a massive

class hierarchy 120 classes each with

various methods inheritance going 10

levels

deep those are some rookie numbers okay

10 levels that's all you got that's all

you got Gil foil the only problem all

the methods were empty I do not

exaggerate here not mostly empty

empty okay okay I like this I like this

there's there's a thing here there's a

thing here that one stumped me for a

while eventually I learned this was in

service of building a structure he could

then use reflection on I oh my oh that's

so beautiful that reflection would let

him create a pipe delimited string whose

structure was completely database driven

but entirely static and he would send

over a socket turns out this was all

eventually sent to uh kellwell kiwi kill

the service that talked to shipping

carriers why did this bug happen

kellwell used nine digigit long numbers

every month someone had disabled the

Cron job that deleted old

orders you know there's something that's

very beautiful about that I'm just going

to be real here for a second this idea

of using reflection to build up a string

and every method was empty it's so

chaotic it's beautiful it's lovely you

know I I mean that's the thing is that

when you see truly novel ways to solve

something there's something about it

that just gets me excited yeah it's

sorcery level this is beautiful in the

same sense as a black hole exactly it's

black hole beautiful it's not that you'd

ever want to be remotely close to a

black hole and get spaghettified into

the universe it's that its beauty is

also why it's so terrifying right it's

its beauty is in its terribleness it is

it is great in the classic sense of the

word great just like in yeah it's just

like a volcano like a volcano is

beautiful it's awesome there you go it's

awesome uh the Beautiful Mess there are

so many more things to tell from this

code base like a team of super senior

developers who were rewriting the whole

thing without shipping any code for 5

years or the red hat Consultants

building one the one database to rule

them all there were so many crazy

corners of this code base so many

reasons why there were a whole teams

dedicated to starting from scratch on

just one bit of its functionality but I

think most important story to tell about

or but I think the most important story

to tell is about Justin's Improvement of

the merchant search page the merchant

search page was the entry point into the

entire application every customer

service rep would get on the phone with

the merchant and type either their ID or

name to find their information that

would land you on a massive page with

all their information the page was

information Debs in the best way full of

any information you could need and any

links you could want to visit but it was

dog slow Justin was the sole senior

developer in my group he was bright

snarky and couldn't care less about the

business he told it like it was didn't

pull any punches and could always solve

problems by himself faster than teams

around him one day Justin got tired of

hearing about how slow the merchant

search page was and went to fix it every

box on the screen became its own endo on

load everything above the fold would

start fetching and as one load in more

requests would come in it took the page

time from minutes to subc Well Done

Justin also kind of wild really like an

npl1 query problem that went

on just absolutely no

respect for the servers let's go let's

go uh two ways to decouple why was

Justin able to do this because the

codebase had no master plan there was no

overarching design the system had to fit

into no expected format for apis no

documented design system no

Architectural Review Board making sure

things were coherent the app was a

complete and utter mess no one could fix

it uh no one could ever fix it so no one

tried to what did we do instead we

carved our only own little world of

sanity this monolithic app and due to

its sheer necessity had grown to be a

microcosm of nice small apps around its

edges each person when tasked with

improving some part of the app would

inevitably give up untangling that web

and find some nice little corner to

build new things and then slowly update

links to point to their nice new stuff

orphaning the old one this may sound

like a mess to you but it was remarkably

enjoyable to work it gone were the

concerns of code duplicate gone were the

concerns of consist consistency gone

were the concerns of extensibility code

was written to serve a use to touch as

little of the area around it as possible

and to be easily

replicatable or replaceable our code was

decoupled because coupling it was simply

harder it's actually kind of nice it's

kind of

sweet it's kind of like a little sweet

ending here this is this is such a

lovely story I I love the fact that this

guy could work in potentially one of the

worst code bases ever created and just

see the positive part of it all I really

love trying to look at things in the

positive light and this is just like the

ultimate expression of being positive

even about something that could be so

awful lovely after in my career sense

I've never had the privilege of working

in such a wonderfully ugly code base

every ugly codebase I've encountered

since has never transcended into uh its

need for consistency perhaps it was

because the codebase had been abandoned

by serious developers long before all

that remained were rag tag interns and

Junior developers or perhaps it was

because there was no layer between those

developers and the users no translations

no requirements Gathering no cards just

you standing at the desk of the customer

service rep ask them how you could make

their life better I missed that direct

connection that fast feedback that lack

of making Grand plans the simple problem

and code connection perhaps it's a

simple it's simply a naive Nostalgia but

just as I find myself laying on the

couch longing to go back to some of the

worst years of my childhood when faced

with yet another Enterprise design

pattern my mind flashes back to that

beautiful horrible code

base I've had something that's kind of

like this in the sense that it was just

like every team had their own little

thing and it was just like a hod Podge

together application and it was

beautiful it wasun to work in I could

get my stuff done and it was virtually

impossible to look at anybody else's

stuff there is something that's very

beautiful about the ability to just

disregard everything and just build your

own

thing there is something beautiful about

that I do love this this was

fantastic thank you Jimmy Miller

beautiful absolutely beautiful the

name I'm I'm I'm a little touched I'm a

little climed right now A Jen