Cost per line of code
jtzecher
I just tried doing a little googling but couldn't come up with the
answer that I'm looking for. Just wondering if anybody had knowledge of
any industry "rule of thumb" estimates on the cost of a writing a line
of code. I recognize that this could easily vary by what language we're
talking about. Specifically, I would be interested in Smalltalk and
JAVA.
On a slightly seperate question, how many lines of JAVA code (on the
average of course) would be equivalent to a line a smalltalk code? I
have been throwing around the number 6 in conversations lately but I
came to realize that I might've just pulled that number out of the air.
Anybody got a handle on a real number?
Many thanks,
Joel Zecher
ke...@compuserve.com
because they are more scarce? Is it getting easier to train up new
Smalltalkers because more people Get OOP these days what with C-- and
Cflat and Joe. Is this indeed what has happened? Joe Schmoe could
brook no grit till he grit the path with it. I have read a note from a
guy who said that a Smalltalk project was converted over to Moe and
they needed twice as many people on it. Course they could get them.
And with Moe there are all these third party tools because you can't
easily make your own. And the thing is busted to begin with anyway.
You can't make Generators in it. And if you can't make Generators in
it what can you do. Lots and lots. Of old style down home country
buffet cooking. But what if you are a sushiman. In Smalltalk you make
your own third party tools. I needed to modify the Dolphin Smalltalk
Packages and I just did it. A little tweek. Just a touch. The source
code was right there. All commented and readable. The Moe pushers say
that home grown third party tools are no good. You need real third
party tools. So you can shore up your broken down system that was
broke when you got it. But don't listen to me. I am a Smalltalker and
a Lisper and am enamoured of things that work. If it doesn't work then
I'm screwed because I can't bang on things random like and just make
them work. just can't do it. isn't part of my nature. It has to work
and work well or I'm screwed. It can be theoretically complex and
that's all right but if it doesn't work. And work all the time every
time, then I'm screwed. I remember back in 1975 the Microsoft C
compiler was so full of bugs they crawled right up your pants legs.
TurboC was better. A lot better. But then Smalltalk came and just
blew them all away. I remember this Modula compiler. The program got
to be a certain size and gzornaplat it locked up. Show's over.
Nothing could be done. Nothing that I could do. But then Smalltalk
came along and just blew them all away. Always clean. Always right
there.
jtzecher
You are preaching to the choir. I agree with all that you have said.
I'm still looking for numbers to use because that is what management
understands best. The non numerical pros and cons of one language over
another are more difficult to convey than quantum theory.
Thank you,
Joel
Alan Knight
Jones and Software Productivity Research (SPR). Doing a google for
"software productivity research language list" yielded the apparently
dead URL http://www.theadvisors.com/langcomparison.htm but whose google
cached version is at
http://www.google.ca/search?
q=software+productivity+research+language+list&sourceid=mozilla-
search&start=0&start=0&ie=utf-8&oe=utf-8&client=firefox-
a&rls=org.mozilla:en-US:official
"jtzecher" <joel....@ge.com> wrote in news:1168293887.398531.239040
@q40g2000cwq.googlegroups.com:
David Buck
Stephan Eggermont
> On a slightly seperate question, how many lines of JAVA code (on the
> average of course) would be equivalent to a line a smalltalk code? I
> have been throwing around the number 6 in conversations lately but I
> came to realize that I might've just pulled that number out of the air.
> Anybody got a handle on a real number?
When creating web applications in Seaside? Somewhere between 100 and 1000...
Stephan
Joachim Tuchel
> When creating web applications in Seaside? Somewhere between 100 and 1000...
While I disagree with these numbers (what's in a line of Java code,
anyway?), in case of web applications the complexity is not coming from
the java code alone but from the large zoo of techniques you need to
handle for a few simple web pages. Seaside can make things very simple
in that respect...
jtzecher
Cesar Rabak
You need to do research in your area about:
1) What's the average hourly rate for Smalltalk and Java programmers;
2) the availability of them for hire;
3) the average productivity in _your problem domain_.
Then you'll have hard data to show to your management.
geoff
a wide range of projects, ST comes out way ahead. With Java, they try to
add some easy things to increase the FP count but Java is still way behind
ST.
An example of a technology with a very low FP count is Lotus Notes
programming, the use of which IBM endorses.
Once again, ST is that red-haired, freckled face kid who gets slapped in the
face since actions do not line up with facts but the actions do line up with
politics.
-g
Eliot Miranda
any hard data/urls for the IBM stuff?
--
The surest sign that intelligent life exists elsewhere in Calvin &
the universe is that none of it has tried to contact us. Hobbes.
--
Eliot ,,,^..^,,, Smalltalk - scene not herd
geoff
No, that info has been buried and IBM sold their smalltalk. Makes sense
though since one can not have a product like smalltalk outshine java. There
still might be LN/Java info, etc.
--g
geoff
1. The data is passed around in spreadsheet form via e-mail as part of a
project review.
2. ST has gotten as high as 60 FP's per 100 man hours, Lotus Notes 30, Java
15 to 20, and PL/1 5.
3. The info has never been posted on the IBM intranet but I am sure it
would be an easy thing for the FP counters to make up a chart and post it if
they wanted to.
-g
Ian Upright
developers, who fully understand how to achieve a tremendous amount of
reuse, end up distilling common patterns and create resuable frameworks and
patters that can make the same functionality 10x or even 100x smaller in
terms of volume of code, than it could take to otherwise (inneficiently)
build code that achieves the same thing. This can especially be the case
when using tools to generate code from models, etc.
Sometimes the 10x or 100x reduction takes real innovation and insight and
radical approaches, and that certainly doesn't come for free.
So at times it may actually be extremely expensive to build a very small
amount of code. However, that very small amount of code could be extremely
powerful and hence could used to run substantially large applications and
functionality. For example, as applications become more meta-data and
data-driven, the amount of code for substantial applications could be nearly
non-existant, with perhaps 95% of application development time building
models, UI, and data that dictates behavior, and 5% developing programming
code specific to that application.
So with such variance, I'm not sure how useful a metric such as this would
be. Perhaps with other programming languages, reuse is not emphasised as
much and so LOC is a more meaningful benchmark. ;)
How many LOC it takes to do the same thing in Smalltalk vs. Java is a far
more meaningful benchmark, but it still can vary from application domain and
also programmer skill, and in my experience could be anywhere from 2x more
LOC in Java than Smalltalk, up to 8x or more. The biggest difference is in
the volume of static type information, casts, declarations, and other
type-related superfluity.
Cheers, Ian
"jtzecher" <joel....@ge.com> wrote:
Bob Nemec
"Ian Upright" <ian-...@upright.net> wrote in message
news:d9q5s214evnor1k3k...@4ax.com...
Chris Uppal
> Refactoring in Smalltalk... a good day is a negative code count ;-)
As in: "Bugger! I just deleted the repository!" ?
;-)
-- chris
Ian Upright
;-)
Ian
"Chris Uppal" <chris...@metagnostic.REMOVE-THIS.org> wrote:
Eliot Miranda
making up a chart is falsifying evidence. If someone were to use an
anonimiser to post the spread sheet it still wouldn't be evidence
(missing proof of the sheet's validity) but it would be better than an
anecdotal report.
Cesar Rabak
> Refactoring in Smalltalk... a good day is a negative code count ;-)
>
the result would be the ammount of (final) locs divided by the time
(effort) employed.
If people starts to measure that way. . .
geoff
have all this data at their fingertips. They generate the performance data
for all the internal projects and somehow, I doubt, they hit the 'delete'
key after a project is over.
-g
glyno...@gmail.com
Is your approach waterfall, hybrid or agile?
Is your approach a cots product to be or SaaS?
What is it your developing and its use?
Is the design government and industry collaboration of or industry?
Physical or logical method?
The list goes on and on. I have found large companies love millions of line of code because the more code the more maintenance, the more maintenance the more you need them, the more you need them the more you DONT grow.
Europeans are big on Lines of CODE - where point of fact, a $1M USD project in 6 months with with 10kLOC may be reasonable based on what it is the customer wants. In pursuing this, you have to consider the PM time, rqts time, analysis time, UAT time etc to have holistic amount - but at the end of the day - no matter what you produce, you have a budget, you have a target, meet it - that simple - if you code to lines vs coding to a common sense approach to provide what is needed, maybe people should consider other professions.
Just a humble opinion and I wish you luck in finding the answer.