MAT/Linux PLC Big Pix (was Open Control Effects on

P

Thread Starter

Paul Jager

Has anyone in the developer camp done an estimate of time to completion of a comprehensive and useable product?

Here's one for starters:

To create a product for specialized use: 6 developers *2000 hours = 12,000 hours

So if you had say 12 programmers donating 100 hours per year - Release date = 2012.


Paul Jager, CEO
www.mnrcan.com
[email protected]
(250)-724-1402
 
J

Joe Jansen/ENGR/HQ/KEMET/US

100 hours per year works out to just less than 2 hours per week. I think your estimate is a bit low. There are times I spend more than that per
day.

Admittidly, my C skills are not exactly the *ahem* strongest in the world, but I would suggest that I am still getting more than 2 hours per week output.

I would point you to more 'real world' examples. ie: look at some of the desktop projects like KDE or Gnome. The man-hours spent on one of those
would dwarf anything that we would be doing in MAT.

The Linux kernel evolution is another one that has 1000's of man-hours being expended, at a rate much higher than 100/year.

I would be curious as to how you arrived at your figures, as maybe we could, acedemically, play around with it a bit.

--Joe Jansen
 
Paul Jager:
> Has anyone in the developer camp done an estimate of time to completion
> of a comprehensive and useable product?

No, not that I know of.

> To create a product for specialized use: 6 developers *2000 hours =
> 12,000 hours

> So if you had say 12 programmers donating 100 hours per year - Release
> date = 2012.

The final result obviously depends a lot on the numbers you pick.

Also, we would hope that the number of contributors will pick up once the
product starts being partially useful - once we reach a point where we can
be used by somebody, hopefully the resulting contribution will make it
useful to somebody else and so on. It should snowball.

Right now its floating point math is pretty good, HMI is ok, but discrete logic is lagging behind, the I/O is limited and the manual is partial.


Jiri
--
Jiri Baum <[email protected]> http://www.csse.monash.edu.au/~jirib
MAT LinuxPLC project --- http://mat.sf.net --- Machine Automation Tools
 
Paul Jager:
> Has anyone in the developer camp done an estimate of time to completion
> of a comprehensive and useable product?

No, not that I know of.

> To create a product for specialized use: 6 developers *2000 hours =
> 12,000 hours

> So if you had say 12 programmers donating 100 hours per year - Release
> date = 2012.

The final result obviously depends a lot on the numbers you pick.

Also, we would hope that the number of contributors will pick up once the
product starts being partially useful - once we reach a point where we can
be used by somebody, hopefully the resulting contribution will make it
useful to somebody else and so on. It should snowball.

Right now its floating point math is pretty good, HMI is ok, but discrete logic is lagging behind, the I/O is limited and the manual is partial.


Jiri
--
Jiri Baum <[email protected]> http://www.csse.monash.edu.au/~jirib
MAT LinuxPLC project --- http://mat.sf.net --- Machine Automation Tools
 
P
From our experience, and others we've seen first draft software development teams are typically ~ 6 people. There are others of course around the project core, but that is the crux. Basically as long as the software is profitable, that team keeps developing, features, patches and releases etc.

The 1year to develop anything meaningful is just a rough guess. I don't think it would be less. If you can put two hours a week into an open source project, that's pretty good.

The industrial business is not main stream, so it takes several years or more of experience to appreciate what goes on inside facilities of the different industries. This further reduces your available "wide area" pool of contributors.

To sum it up, I think there is a shortage of available, qualified labor.

Paul Jager, CEO
www.mnrcan.com
[email protected]
(250)-724-1402
 
J

Joe Jansen/ENGR/HQ/KEMET/US

6 active developers on a team. OK, I can go along with that.....

1 year to project completion. If you define 1 year as 2000 hours per developer and not as 12 months, I can agree. LPLC has already passed 12 months, which is why I make the distinction.

2 hours per week is still significantly less than what I am able to contribute. I typically get 1.5 to 2 hours per day. Maybe my life is just empty..... How depressing.... ;^}

--Joe Jansen
 
Paul Jager:
> From our experience, and others we've seen first draft software
> development teams are typically ~ 6 people.
...
> The 1year to develop anything meaningful is just a rough guess.
...

OK, makes sense. A year for a first draft sounds a bit long, but I guess that depends on the size. And after all we've been taking a couple of years already (the project started at the beginning of 2000, and the first code was checked into the CVS in May that year - what became the Global Map).

> The industrial business is not main stream, so it takes several years or
> more of experience to appreciate what goes on inside facilities of the
> different industries. This further reduces your available "wide area"
> pool of contributors.

On the other hand, if we get a contributor from inside such a facility, then such a person is (ideally) qualified to appreciate what goes on there. Presumably, this'll only work if we're already close - but it should work for the ``last mile''.

In some ways, this is a mile that the large commercial houses cannot hope to match, because any feature request has to filter through many different hands before it gets to anyone in a position to do something about it. For one thing, this results in inevitable distortions; more importantly, though, it takes so long that it will not be available for the project currently at hand anyway, so the would-be requestor doesn't bother.


Jiri
--
Jiri Baum <[email protected]> http://www.csse.monash.edu.au/~jirib
MAT LinuxPLC project --- http://mat.sf.net --- Machine Automation Tools
 
Top