How big is big?

dougal85

I often hear people talking about projects being small or big. saying things like 'in a small project I wouldnt really worry about that.'

However, how big would you say is big?

1000 lines? 5000? 10000?

I guess it depends on the person really. Anyway, i'm just a little bored and curious at the same time.

NogDog

While 10,000 lines of code would probably be considered "big" for any one person working on their own, it's pretty tiny compared to really large projects. Years ago I worked on the ENWGS program which topped out around 1.25 million lines of code. This of course evolved over quite a few years, with anywhere from a dozen to 40 or so programmers working on it at any given time.

dougal85

wow thats incredible. quite impressive. maybe i should start a thread and see who's got the most lines of code for a solo project!

Weedpacket

NogDog wrote:
1.25 million lines of code.

In 2001 the Linux kernel was pushing two and a half million lines, and Mozilla was not far behind. But they probably had a slightly larger developer base 🙂

Y'know, NogDog; after seeing that I'm reminded of those estimates back in the '80s about the expected size of the codebase that would be needed to properly implement the Strategic Defense Initiative. Ten million lines was regarded as low. If I'd known about ENWGS at the time I would have (a) been surprised to, and (b) agreed that ten million is WAY too low.

NogDog

I should probably point out here that I did not personally write any of those 1.25 million lines, as I was a test engineer on that project. I learned how to read the PL1 code (and later the Ada code) and use the debuggers to help find out where problems originated, but I didn't do any of the actual programming. It was definitely the most interesting 9 years of employment for me so far, though. 🙂

dream_scape

dougal85 wrote:
maybe i should start a thread and see who's got the most lines of code for a solo project!

According to an SLOC Count utility, the FlexieCommerce project that I'm working on [by myself] currently has 41,728 lines of code... It's in the middle of a large re-factoring though at the moment, so there is currently a lot of redundancy and deprecated code that will be gone at the end of the re-factoring.

sneakyimp

Wikipedia wrote:
Finally, on March 14, 1994, Linux 1.0.0 was released, with 176,250 lines of code. In March 1995, Linux 1.2.0 was released (310,950 lines of code).

Funny how these things grow. Wikipedia also says that linux v 0.1 had only about 10,000 lines.

When lines of php script get counted, do you:
1) count white space lines?
2) count html lines?

the 2nd question might seem retarded, but seems to me that a line like this should maybe be counted:

<td><input type="text" value="<?=empty($_POST['foo']) ? '' : htmlspecialchars($_POST['foo']) ?>" name="foo"></td>

or maybe javascript code...

dougal85

lol... there is no really easy way to count anything besides, i could just lie to you anyway 😛

It's just interesting to see what people say about it.

I read XP is 10 million... but i cant remember where.

Piranha

dougal85 wrote:
I read XP is 10 million...

And of those 9 million contains bugs... :p

Unfortunatley I use XP, haven't gotten around to learn Linux yet.

sneakyimp

you might want to start with knoppix:

http://www.knoppix.org/

it's got a GUI which looks a lot like windows.

Weedpacket

sneakyimp wrote:
1) count white space lines?
2) count html lines?

The SLOC metric that dream.scape mentioned and I used is more a rule of thumb measure; good to within an order of magnitude or so, and a lot easier to get at than more objective measures. (Random statistic I heard somewhere; typical software contains bugs at the rate of 5-15 per thousand lines - depending on how you count lines.)

Legend has it there are development houses out there that pay their programmers by how many lines of code they produce. Me, I don't believe that such companies would be able to survive long enough to ship. But then, I tend to overestimate the intelligence of the average person.

dougal85 wrote:
I read XP is 10 million... but i cant remember where.

Another perspective: look at the size of the Windows API. Windows 1.0 had about 400 functions and Windows 95 more than a thousand. The .NET 2.0 framework has about 5000 public classes, with 45,000 public methods and 15,000 public properties (not counting those that are just inherited from parents), all of which have to be implemented.

dalecosp

Might be worth mentioning that big to write and big to maintain can be rather different things, as well.

Weedpacket

dalecosp wrote:
Might be worth mentioning that big to write and big to maintain can be rather different things, as well.

I wonder if there's a software metric that measures how much the programmer has to keep track of as a function of the size of the change being made? With buried away in it an expression like "number of lines of code potentially affected"/"number of lines actually changed"?

NogDog

OMG! I'm having flashbacks to Function Point Analysis!!!

(Is that still a big buzz-word in software development?)

goldbug

Weedpacket wrote:
The SLOC metricLegend has it there are development houses out there that pay their programmers by how many lines of code they produce. Me, I don't believe that such companies would be able to survive long enough to ship. But then, I tend to overestimate the intelligence of the average person.

And using such a simple and easily-gamed metric is a poor way to judge a programmers worth. Something that may take 10 condensed lines of solid code could easily be made into 100lines by the person payed per-line.

And "lots of lines" really is dependent on what's counted--code,comments,whitespace,etc--but still it's too vague to be used as a quality metric. If you assume the opposite of the pay-per-lines stance--more concise code is better--then you get crunched up code that is impossible to maintain. Somewhere in-between extra-bloaty and extra-tiny code is probably some happy zone where programmers might want to be.

For statistical purposes, I just ran an app I wrote the first 2/3 of 2006 (roughly) through K-LOC (I have doubts about it's accuracy and whitespace exclusion)....it came out to around 365,000 lines (again, I'm looking at that number with a skeptical eye)....and that's not including architecting/diagramming either. But again, that's not a good judge of either the quality of the code or the productivity of the programmer. Keep in mind in that same 8 months I also did 100% of the design/imagery/etc for the application--from scratch.....you can't measure that input by counting lines.

So if you want to count lines of code, go ahead, but remember it's a meaningless metric, because it it doesn't include quality, externalities, or even language into the mix (we all know you can take a berjillion lines of Java and crush it into one extremely cryptic line of Perl.).

dougal85

Weedpacket wrote:
Another perspective: look at the size of the Windows API. Windows 1.0 had about 400 functions and Windows 95 more than a thousand. The .NET 2.0 framework has about 5000 public classes, with 45,000 public methods and 15,000 public properties (not counting those that are just inherited from parents), all of which have to be implemented.

Do you just remember all those numbers? 🙂

goldbug wrote:
And using such a simple and easily-gamed metric is a poor way to judge a programmers worth. Something that may take 10 condensed lines of solid code could easily be made into 100lines by the person payed per-line.

lol, yeah if i was getting paid by line you could change allot of things to easily make it grow; ie...


$something = ($condition)? TRUE : FALSE;

// could be

if($condition){
$something = TRUE;
}else{
$something = FALSE;
}

That just gave me a thought, imagine making Rube Goldberg styled php applications, make the most complicated application to do the smallest thing. lol

goldbug

dougal85 wrote:
lol, yeah if i was getting paid by line you could change allot of things to easily make it grow; ie...
$something = ($condition)? TRUE : FALSE;

// could be

if($condition){
$something = TRUE;
}else{
$something = FALSE;
}

I'd pay more to see:

$something = ($condition);

In that example 😉

dougal85

yeah yeah, well done! lol. but the example still makes my point 😛

Piranha

But why be moderest? You can expand that code further:

if($condition)
{
    $something = TRUE;
}
else
{
    $something = FALSE;
}

8 lines of code instead of 1. Basically it would be easy to earn money that way. Especially doing totally meaningless things like always use SELECT * in queries, always counting the rows and get everything into variables.

dougal85

edit scrap what i wrote the first time. lol. i must have been tripping.

lol, ok... how complicated can we make that statement i wonder? although, its easy since php doesnt take a new line as a new line, hence the need for ;

if
($condition)
{
    $something 
    = TRUE;
}
else
{
    $something 
    = FALSE;
}