[rolf.kaiser]

Hello Members,
I would appreciate of you can share your thoughts on How can someone defend the claim of using in house coding for industrial application over commercial coding, Any guide will be appreciated.

thanks in advance
rolf

[cfdnewbie]

Depends on the problem. Inhouse can of course be tailored to your need, extended, debugged, comes free of charge...l

[vikas171092]

Open source code comes handy for academic research only... like someone can't rely on the results of open source CFD software for analysis of a jet engine.

[cfdnewbie]

Originally Posted by vikas171092

Open source code comes handy for academic research only... like someone can't rely on the results of open source CFD software for analysis of a jet engine.

I'm not sure if I understand your last sentence, but are you saying that open source codes can't be trusted?
In my experience, open source codes offer a great deal of trustworthiness, a lot more than commercial codes. Why? because you can do all the testing you like yourself, you can look "under the hood" and see exactly what's going on....

Commercial software is often just a black box, with no means to look at the actual code, do meaningful testing etc.

[vikas171092]

Lets understand it by an example:

I want to design a military aircraft, so i used say X commercial product, now these are certified companies, and if anything goes wrong the company is to be blamed. So that much trust a commercial software provides. Who will be blamed if open source software is used. I agree open source is excellent as far as learning is concerned.

Hope now u have clear picture, what i mean to say..

[cfdnewbie]

Originally Posted by vikas171092

Lets understand it by an example:

I want to design a military aircraft, so i used say X commercial product, now these are certified companies, and if anything goes wrong the company is to be blamed. So that much trust a commercial software provides. Who will be blamed if open source software is used. I agree open source is excellent as far as learning is concerned.

Hope now u have clear picture, what i mean to say..

hm, I get your example, but I'm not sure if that's how it would go in the real world - I'm no law expert, though. If you licence/buy a commercial product and use it to produce results (answers to an engineering problem), I would assume that you (the one who created the results) is responsible for them, not the manufacturer of the code. I assume that it would be your responsibility as a professional engineer to determine whether the results you got with the software make sense or not. I doubt that you can sue software companies for faulty software - can you sue microsoft because windows crashes? Don't think so.

But of course, if you do use your own/ open source software, the only one responsible in any case would be you - that can be a good or a bad thing, depending on your standpoint.

From my experience, you have a lot more control over open source software than over commercial one....and that means that you can have a much higher confidence in your own code than the black box.

There are not few large engineering companies that nowadays push towards open source software - I'm sure that they have their legal departments, and they wouldn't touch open source if it meant any danger of lawsuits to them - because they can get better results with a lot less investment. From my experience, commercial code makers will lose a considerable market share in the next decade or so.

[vikas171092]

I am sorry, i don't agree with you. As i am currently doing internship in Ansys Inc. so i have a clear picture about an open source and a commercial CFD package.

[cfdnewbie]

No problem, everyone has his own perspective, but of course Ansys will have its specific way of looking at the issue.

There will always be a market for commerical products with nice GUIs and fast support, as there will be a grwoing market for open source codes. One question Ansys and co haven't answered yet is: If I want to run your code on 10000 or 100000 cores, do I have to buy 100000 licences? Does your code scale beyond 100 or 200 cores? I have recently seen the scaling of CFX on up to 300 cores, and over 200, the results look really sad....

[cfdnewbie]

You are giving up?

[sail]

Originally Posted by vikas171092

Lets understand it by an example:

I want to design a military aircraft, so i used say X commercial product, now these are certified companies, and if anything goes wrong the company is to be blamed. So that much trust a commercial software provides. Who will be blamed if open source software is used. I agree open source is excellent as far as learning is concerned.

Hope now u have clear picture, what i mean to say..

actually Ansys' licence says exactely the opposite.

commercial software is excellent. usually reliable, widely tested, and the assistance given from the company (if timely ) is an enourmus advantage.

but...

you have to put your faith in the product you don't know how it works, knowing that if somethig goes wrong you are the only responsable in front of the client.

[vikas171092]

I didn't understand what you said .

[cfdnewbie]

Originally Posted by vikas171092

I didn't understand what you said .

which part?

[cfdnewbie]

Originally Posted by sail

actually Ansys' licence says exactely the opposite.

ah, good to know, I thought it had to be that way.... after all, cfd software should only be used by trained professionals, and it is not an infallible tool with a valid solution for every problem.

I agree that commercial tools can be very helpful. Another plus is the support - but truthfully, some of my worst students end up as end user support at these companies, so I personally wouldn't call there

But be that as it may - there's a market for both commercial and open codes, and the market for open codes is growing.

[vikas171092]

that license stuff !!

[cfdnewbie]

Ok, so let's say a company wants to run the CFD code on 256 cpus in parallel. Often, they then have to buy the software 256 times, like buying 256 copies of windows for all cpus. How much does that cost?! A lot!
And what about if they want to run the code on 100000 Cpus?

With open source software, there's no limit on how many copies of the software you need, with commercial software, (usually) you pay per copy.

(The parallel performance is another issue alltogether, I have yet to see a commercial product that scales beyond 256)

[sail]

Originally Posted by vikas171092

that license stuff !!

from ansys website, legal stuff... http://www.ansys.com/staticassets/AN...st-22-2011.pdf

Licensor, its Affiliates, Channel Partners and suppliers do not warrant the accuracy or the applicability of the results
obtained from the use of the Program(s) or the Manual. No other documents or oral conversations, statements or
representations will be offered by Licensee as evidence to explain, expand, alter, add to or invalidate the express
warranties set forth above.

In the event that Program(s) fails to perform in all material respects as warranted in this Agreement, Licensee's sole
remedy will be for Licensor, at Licensor's option, to:
(i) Provide a correction or work-around to correct the breach; or
(ii) Modify the Program(s) to conform substantially to the Manual; or
(iii) If the Manual is in error, modify the Manual to accurately reflect the Program(s)'s intended functionality and actual
operation; or
(iv) Terminate the license for that Program(s) and/or this Agreement and require Licensee to return the Program(s) to
Licensor, in which event Licensor will refund to Licen...

it's almost a standard software license agreement, so you use the prorduct at your own risk.

I will repeat myself, I have used with satisfaction commercial software, and they have some advantages over their open source counterparts. But limitation of responsability is not amongst them.

[cfdnewbie]

Originally Posted by sail

But limitation of responsability is not amongst them.

Which is a good thing, otherwise, nobody would dare to sell these products any more. plus, everbody using this software should be a well-trained professional, who knows what he is doing or has at least the resources available to read it up somewhere.....

I guess that's why I don't look commercial software: The user can have no knowledge about what's going on, make the stupidest mistakes, but the code will usually produce some nice pictures...it's like landing on autopilot, it works allright, until something unforeseen comes along...then: Booom!

I'm not saying that all users of commercial products are totally uninformed, but if you browse the forums here, you will get the impression that many many people are told by their boss to solve a problem with the new shiny CFD tool, and haven't had any training and have never taken a class in PDEs....