How Intuit Manages 10 Million Lines of Code

CowboyRobot writes "Intuit launched QuickBooks in 1992, and it has grown into the best-selling retail software for small-business accounting worldwide. QuickBooks is available on multiple platforms with different feature sets (Pro, Premier, Enterprise), in specialized editions (accounting, contracting, etc.), is available on CD or via subscription, and is offered in localized versions for the U.S., Canada, and the U.K. How they manage so many builds is a case-study for large scale programming. 'The Windows version is about 80,000 source files, 10+ million lines of C++ code plus a little C# for the .NET parts. Plus help files, tax tables, files defining local accounting rules, tax and other government reporting forms, upgrade offers — a lot of files. Every customer gets the full version. Specific feature sets are turned on and off with the license key.' And the lessons are not just technical. 'One surprising lesson is that small teams work, even for very large codebases — especially, Burt says, in sustaining an entrepreneurial, creative culture.'"

Comments

-2000 Lines Of Code

[Osgeld]

n early 1982, the Lisa software team was trying to buckle down for the big push to ship the software within the next six months. Some of the managers decided that it would be a good idea to track the progress of each individual engineer in terms of the amount of code that they wrote from week to week. They devised a form that each engineer was required to submit every Friday, which included a field for the number of lines of code that were written that week.

Bill Atkinson, the author of Quickdraw and the main user interface designer, who was by far the most important Lisa implementor, thought that lines of code was a silly measure of software productivity. He thought his goal was to write as small and fast a program as possible, and that the lines of code metric only encouraged writing sloppy, bloated, broken code.

He recently was working on optimizing Quickdraw's region calculation machinery, and had completely rewritten the region engine using a simpler, more general algorithm which, after some tweaking, made region operations almost six times faster. As a by-product, the rewrite also saved around 2,000 lines of code.

He was just putting the finishing touches on the optimization when it was time to fill out the management form for the first time. When he got to the lines of code part, he thought about it for a second, and then wrote in the number: -2000.

I'm not sure how the managers reacted to that, but I do know that after a couple more weeks, they stopped asking Bill to fill out the form, and he gladly complied.

http://folklore.org/StoryView.py?story=Negative_2000_Lines_Of_Code.txt

point is, just cause you can manage it, doesnt mean 10,000,000 lines of code is really something to brag about, especially for something that feels as cheap as quickbooks (though it does a ok job if your accountant cant use excel and must have things that visually represent checks)

Re:-2000 Lines Of Code

[Lev13than]

point is, just cause you can manage it, doesnt mean 10,000,000 lines of code is really something to brag about, especially for something that feels as cheap as quickbooks (though it does a ok job if your accountant cant use excel and must have things that visually represent checks)

If your accountant is using Excel to run your books that means it's time to get a new accountant.

Re:-2000 Lines Of Code

[Osgeld]

excel is a tool not a solution

quickbooks is more of a solution, but like the the 150 piece toolkit for 19.95 at harbor freight, it may do the job just fine, it may snap under the load

Re:-2000 Lines Of Code

[Sir_Sri]

doesnt mean 10,000,000 lines of code is really something to brag about

No, but managing 10 million lines of code successfully is something to brag about.

However it got there, it's there now, and trying to figure out what to do with it, in a way that gets customers their product, teams their appropriate access, doesn't break every time a new intern looks at it, and can track changes back to individual employees so you can evaluate them somehow.

I'm not saying this isn't a problem other people haven't solved - certainly they have, but different solutions may offer some pieces of unique insights.

Re:-2000 Lines Of Code

[ciderbrew]

because tax laws are mental and getting a database to do that would drive it insane. I had to spec out a tiny bit of function with an accountant and the amount of exceptions to the rule made it pointless to have a rule.

Re:Bah.

[russotto]

Given that they haven't managed to come out with a native version of Quicken for Mac in over 6 years, I suspect that they kept that code on a napkin... then lost the napkin.

Re:Bah.

[JoeMerchant]

Real programmers kept their code on punch cards.

Or toggle patterns written on a napkin.

FTFY

My college took out the card punching stations the semester before I took Fortran, though they still had the vacuum drum reader and used it occasionally. And, of course, the CRT terminals were inputting card format - first six characters for line number, 78 characters per line max.

The fossilized prof told stories of entering assembly op codes, in octal, with rotary dial interfaces.

Now, get off my... um, I forget, but get off it, NOW!

I thought I disabled ads.

[WiiVault]

Oh and as most anybody who uses QB often knows- it does suck quite mighty and gets slower even as PCs get faster. Piles of code isn't always a good thing, sometimes it just means you have a bloated heifer with lazy programming and no real improvements in half a decade or more.

Re:I thought I disabled ads.

[PhunkySchtuff]

As a QuickBooks user, who prefers it to MYOB at least, I ask this question in all honesty:

What else is out there for small businesses that's better than QuickBooks and isn't a cloud-based service?

Re:I thought I disabled ads.

[c0lo]

Piles of code is never a good thing

FTFY.

for anyone that may need an explanation - does anyone? - either the code is necessary (and, thus, won't be thrown into a pile, but carefully and continuously polished and made as small as possible) or it's not - in this case, it's truly a liability.

Re:I thought I disabled ads.

[rueger]

Gnucash is most assuredly not a replacement for Quickbooks, if for no other reason that ta some point you need to interface with a real accountant, and Handing over your files from Quickbooks or Simply Accounting makes that MUCH easier.

Re:I thought I disabled ads.

[afidel]

Sage Peachtree is the main competitor in the self hosted small business accounting realm. To be honest if my dad wasn't so opposed to a hosted solution I would have had him go that way as even with 5 users the cost difference once you add up the server and licenses wasn't that great over five years.

As far as QB goes, it seems to finally be ok. Up until 2007 it still couldn't run as a limited rights user and required short printer names both of which caused me serious pain as I was the IT department for a midsized accounting firm and getting QB running in Citrix was NOT fun. We had premier enterprise support and their answer to the LRU issue was to grant Everyone full rights to HK_Classes_Root (NOT a good idea). I eventually figured out what keys the user needed access to and published the solution online to help others but their spaghetti code was so bad even then that nobody could actually tell what it was doing to give me the right answer.

Re:I thought I disabled ads.

[ekimd]

Bologna. I've been using Gnucash for years on a sizable business with no problem. I'm also very familiar with QuickBooks (I worked as a staff accountant for several years). Sure, there's a lot that quickbooks can do that Gnucash can't, but there's nothing fundamentally missing from Gnucash to prevent someone from using it for business purposes. If FOSS means anything to you, Gnucash is the only way to go.

Re:I thought I disabled ads.

[TemporalBeing]

I disagree. As long as your books are kept correctly, the accountant only needs three things:

1) Profit and Loss 2) Balance Sheet 3) Depreciation Schedule

Gnucash does all of those. Print them out. Hand it to your account. And you're done.

As a small business owner using GnuCash, it's not quite that simple. Accountants like using tools like QuickBooks that they are very familiar with, and they don't want to have to re-enter all the data. So printing out the information is useless for them. You really need to export it to a format that both support.

Sadly, GnuCash does not support exporting Quicken/QuickBooks formats. It will important them with some extra file filters, but no export capability. My solution is to export to CVS which QuickBooks can import as well. Haven't tried it yet, but my accountant & I will be working through it when the time comes.

You mean...

[stanlyb]

That paying well to a small team, and treating them like a people, not commodity, is actually worth the money!!!!
Oh, noooo, no, what about all the HR screams that this is not a sound business???

If it takes 20 million lines of code

[codepunk]

If it takes 20 million lines of code to do ones taxes there is simply something very wrong with the process.

Re:If it takes 20 million lines of code

[Capt. Skinny]

I'd be willing to bet... that a competent group of people could implement QB in under a quarter of the lines of code

Agreed. But then the business aspect comes into play: can they market it well enough to displace the market leader? There's plenty of crap out there that makes money despite the fact that better options exist. If it was as easy as just competent programming, someone would have done it already.

Re:If it takes 20 million lines of code

[kalpaha]

That may be true of most any software, but in almost no case is it wise to do so. Read: http://www.joelonsoftware.com/articles/fog0000000069.html

Re:If it takes 20 million lines of code

[oatworm]

It's not terribly difficult from a technical perspective, but there are a few caveats to keep in mind:
- Everybody and their mother (at least in accounting-related clerical work) knows Quickbooks. Whatever you come up with would have to be similar enough to justify the training expense.
- Intuit really does spend a lot of effort keeping track of various local, state, and federal regulations, at least in the US, and applying them to their software. That's not cheap or easy.
- Since Quickbooks is something of an "industry standard", it's possible to share Quickbooks files among necessary individuals (outsourced accountants and the like) and know that the books are getting from point A to point B. Not everyone has a copy of, say, Peachtree lying around, to say nothing about GnuCash or anything else.
- Accounting is generally not something that businesses start "experimenting" with. Predictable and supported are what they're looking for. Given a choice between a technically superior product from a company that just received angel capital last week and a predictably wrong product supported by a company that's been selling and supporting accounting software for 30 years, most businesses will go the safe route and buy the technically weaker package. It's actually pretty rational if you think about it; switching accounting packages is not trivial by any stretch of the imagination, so picking the product from the company with proven staying power makes a lot of sense.

Personally, I think Quickbooks is kind of the Microsoft Access of the accounting world - oh yes, there are better, far more stable tools out there, but too many people know Access and its quirks for all but one or two of them to catch on in any meaningful sense. That's inertia for you.

Re:If it takes 20 million lines of code

[gmhowell]

I'd be willing to bet a few billion dollars that a competent group of people could implement QB in under a quarter of the lines of code while both improving the quality and making it significantly faster in every regard.

So then why has nobody come up with a better product, then, Mr. Billionaire?

He said 'willing', not 'able'.

Re:If it takes 20 million lines of code

[clintp]

disclaimer: I work in the financial software field in a business very similar to Intuit.

Every time this comes up Slashdot is inundated with lots of comments about how easy this should be. It's not. Get over it.

The problem with financial software is that because of regulatory constraints and tax laws it's nearly impossible to start from scratch. Tax laws are vast. You just won't believe how vastly, hugely, mind- bogglingly vast they are. I mean, you may think it's complicated to write an operating system, but that's just peanuts to tax law.

The secondary problems for payroll compliance and tax law is that they change (with alarming frequency). You probably don't know this because you're a single case, but on the federal and state level hundreds of changes happen each quarter for something like payroll taxes. Almost none of them apply to you. But every one of them applies to someone, and it's something you've got to get right. Oh yes, and let's not forget the thousands of cities, counties, and school districts that all have their own compliance.

In software development, you're left with a few choices in this matter. Pare down the problem, scale up the development, or adopt a progressive model and hope you don't go bankrupt waiting to ship while keeping up with the changes along the way.

If you pare down the problem you've limited your market share. Deciding that you'll only handle, say, Federal filings means that your customers are left on their own for state compliance. It's a niche, sure, but a really small one. That's how every one of the companies you've heard of doing tax filing or payroll software started. And then you work really hard to make your company more and more relevant and scaling up quickly.

For option two, development can't scale this large from scratch. Brooks's Law (in this case) means the entire thing will collapse into a black hole of non-shipping code long before you've made it out of Year 1. And it's got to be nearly perfect -- people will tolerate a certain number of bugs in their games, but getting penalties from the IRS will keep your customers from coming back.

So you want to tackle federal and state taxes (no local) with your launch product? Expect to take at least a couple of years for this, and line up the appropriate capital to keep you in business meanwhile. Plan on using the last 25% of your development and test cycle bringing the code up to compliance with the regulations that have shifted while you were mucking around building the product. (i.e. if you can write and test it in 18 months, expect to ship in 2 years because you'll need the last 6 months just to catch up). That number is from experience, folks. Just try to find a venture capitalist that'll fund this kind of effort into a market that already has a few large players (Microsoft Dynamics, Intuit, Solomon) and dozens of small ones for a product that's just marginally better. Just try.

"I'm a web developer! I deal with shifting standards all the time!" No, not like this. There are thousands of standards and they move as often as quarterly and they're all published separately by different entities that don't interact at all. Screwing up browser standards in a web application usually means something doesn't display right and you find out right away; doing the same in payroll taxes means large penalties vastly disproportionate to the size of your mistake which you don't hear about until the letter arrives months later.

"I'll build clever databases!" "Properly designed rule sets will be my silver bullet!". Nope, think again. You're working against an adversarial opponent: politicians. Politicians need to raise taxes without looking like their raising taxes. Their rule sets can have unlimited complexity in their efforts to tax you while not looking like they're taxing you. They do just make shit up. If Congress wants to have an employer payroll tax for companies over 20 employees that have more than

there's no good competitor

[Skynyrd]

My last job involved many, many frustrating hours with QuickBooks - every week. It's a steaming pile of crap. There were so many basic things missing, like decent reporting, that it was a total joke. I could go on for many long rants about how much I hate the software. We wanted to integrate Fishbowl, so we could do some trick inventory manipulation, but it wasn't implemented before I quit. They are still working on it.

In my new job, I get the daily reports from our 5 stores. QB failures are mentioned daily. I am so happy to be in an entirely different department.

I really wish there was competition for QB. I think it's a fine platform if you are a very small business with a limited product line. Get complicated, and it fails.

Re:Bah.

[WiiVault]

Mac user or not Quicken like Intuit is a joke of a "pro" software company. When Symantec looks to hire people who can pack more useless lines of CPU hogging code they must hire from these guys. How this company has kept it's niche is beyond me.

You can survive

[symbolset]

Quicken and Quickbooks is the only application I know of to have survived a full-on Microsoft assault on their business. Microsoft Money has folded. It's something to be proud of, I guess - for now.

Maybe the small business standard...but

[lilfields]

Quickbooks maybe the industry standard, but it's also a giant jumbled up mess. Our company's quickbooks were down all day due to corrupted files leading to unsequenced invoices. I've never meet someone really happy with Quickbooks...it's just accounting software is really hard to develop, so it doesn't have many strong competitors.

Re:Maybe the small business standard...but

[DigiShaman]

This!

I support Quickbooks as a consultant/sysadmin for SMBs. One of my biggest gripes about QB is that if the client PCs arn't configured correctly, it will attempt to take over the role of sharing out the file while the actual file is still on the server. And it's compounded by the fact Windows XP doesn't route data over the ethernet as a priority over WiFi connection. It can get ugly in the office real quick over "who has the file open". The proper way is to install over the server side hosting agent on the server, and clients arn't configured to host. Although some users get confused over the concept of "hosting" vs "multi-user mode". Bah!!! Curses!

The other major problem is that Quickbooks accounting administrators are not performing backups. Sure, Windows Backup or Backup Exec may be capturing the file. But that's not the same thing. The transaction log will not flush until after Quickbooks itself is allowed to perform a full backup with verification. Sometimes if you wait too long, the entire damn thing gets corrupted. In fact, one of my clients is facing this very problem. She wants to upgrade to the latest version of QB. But until the data can be repaired (if possible) via special services at Intuit, she has no safe way of converting the file without it either bombing out or causing even more corruption.

Re:Was Intuit important in the past or something?

[DogDude]

Intuit makes the most widely used accounting software on the planet. They have significance, but you're just clueless.

Hope they handle their code better than customers

[dotancohen]

The last time that I tried to contact Intuit about Quicken on Linux was a mess:
http://dotancohen.com/eng/quicken_on_linux.html

I certainly hope that they handle their code better than they handle their customers.

Re:Bah.

[DogDude]

Quickbooks and Quicken are similar only in that they are owned by the same company. They are not related in any other way. It sounds like you're giving too much weight to the "Quick" part of the names.

Re:Was Intuit important in the past or something?

[RulerOf]

And, even they had some signficance sometime or somewhere, why should I care about how they manage lines of ancient code?

Because they roll up the money they make on re-selling the same code base year after year, insert the money in their nostrils, then finally, they separate out single lines of code and snort them.

That's why.

Re:No offense, but that doesn't sound like a lot

[cecom]

Yeah, right, you wrote, debugged and tested 500 lines per hour for 800 hours... We believe you. Now take your pills, your straitjacket, go back to your room and let the adults talk.

Re:No offense, but that doesn't sound like a lot

[CodeBuster]

Just as a curiosity, why do you have a single file of more than 30K lines? Isn't that way over the top?

Not if the entire program is one enormous switch statement in the main() function.

Re:Bah.

[shugah]

People who own Macs don't understand technical geeky things like numbers and are way too cool for accounting.

Re:Bah.

[gmhowell]

People who own Macs don't understand technical geeky things like numbers and are way too cool for accounting.

Considering we overpay for everything and buy new iShinies every 3-6 months, I'd say keeping track of our finances is even more important for us than for you bottom feeding 99%ers and your 'PCs'.

Support is the other question...

[bradley13]

Maybe Intuit does feed their programmers well. Now if they would just hire some decent support people. We used to use a multi-currency version of QuickBooks, and the errors were simply astounding in their stupidity. As one, minor example: it was completely clear that some conversions were handled using floating point numbers, with the inevitable decimal rounding errors. At the end of the year, we had huge currency conversion errors - far beyond what even floating-point errors could explain - and essentially impossible to justify to the tax authorities.

There were lots of other problems as well. Trying to report errors was like talking to a black hole. The one time we really needed some real-time support, we spent ages on hold, or going around in circles with different people, and wound up getting no useful help at all. In the end, the only solution was to go back to a single-currency system, and deal with currency outside of QuickBooks.

Still, as bad as our experiences with QuickBooks have been, there's nothing better in the same price category.

Re:Bah.

[JDG1980]

Yeah right. Like they can't afford to hire some Objective-C programmers. As I said before, pretty disgusting for a company who owes its very EXISTENCE to Apple.

It's absurd to say that they owe their very existence to Apple. Does everyone who develops software for Windows "owe their very existence" to Microsoft?

If they thought they could make a profit by porting their software to OSX, they probably would have done it. But large companies generally don't act for purely sentimental reasons.

Re:Bah.

[gtall]

Meds not working like they used to any longer? I know how you feel, try the little blue ones.

Re:Bah.

[hawkinspeter]

They probably do owe their continued existence to Microsoft since their 1997 deal ($150 million investment).

Re:Bah.

[AngryDeuce]

Does that mean Apple owes its existence to Microsoft?

I just felt a great disturbance in the Apple fandom, as if millions of fanboys suddenly cried out in terror and were suddenly silenced.

Re:Bah.

[datavirtue]

No one has made a move into the niche. It is like Windows, Quickbooks is the dominate software and no other company thinks it is worth trying to penetrate. On CNET you can see reviews of their free bait offerings and they are not good--many, many rants. They are ripe for the picking; their software is bloated to hell, expensive, and the users hate it. It is very costly for a small business or start-up to purchase QB and there are no breaks or decent entry points. You pay the QB tax or you don't play, period. On top of this, there is nothing quick and easy about it, and out of the box it makes your company look amateurish.

Re:Bah.

[Sebastopol]

What's the alternative? And don't give me some unix freeware crap, I've tried several, they are crap. Microsoft Money was the only viable alternative, but the conversion process sucked ass. I have 18 years of quicken data that I'm not about to lose, and until there's a real alternative for consumers (i.e., $100 price point), I'm not leaving, regardless of the weird interface, and periodic interface bugs. At least the database is rock solid. (runs to knock on some wood)