Linux V embedded MS. I need a paper that refutes this:

If you want to find a report preferably supporting Linux, someone should fire you :-) If the only tool you have is a hammer, everything will look like a nail.

Just look at the problem at hand, the future of the product (?) and the abilities of the people that have to solve the problem. If you base a choice of the OS on your particular situation, it is almost impossible to make a wrong choice. Reports are just more fud.

--
Reply to nico@nctdevpuntnl (punt=.)
Bedrijven en winkels vindt U op www.adresboekje.nl

What, exactly, did you want refuted? It's naive to think that Linux would

*always* be the best solution. Keep in mind that that report lives out plenty of other commercially successfully embedded OSes such as the Palm OS, Symbian, vxWorks, QNX, etc.

That being said... to me it's pretty obvious that Microsoft has some of the best support out there, and if you goal is time to market it's probably the way to go. On the other hand, *if you let me choose my programmers*, I think I can get you to market in the same period of time with Linux as with Windows. But that's a big "if"...

Look at what the big test equipment guys do... for their mid- and high-level products, HP and Tektronix use Windows. For the low-end stuff, they use Linux. The products at the low end don't provide the margin needed for Windows, yet they're in a market they've been in for awhile so they (presumably) have plenty of good Linux guys around and a workable development system in place.

Under what circumstances? You'd need to offer a shit load of specifications to make a sane decision.

I am in a huge fight with management regarding the design of a 100k+ product. I am against using an OS altogether. Regardless, i need papers, that is what management want to see.

It is at a high level design stage, so specs are pretty light at the moment. I dont intend this to be MS V Linux deabate.

I welcome crosspost to relevant groups.

PS. I intend to win this fight!!! Please help.

try a post to comp.arch.embedded, and maybe an email to Jack Gannsle might be informative

martin

The history of Embedded Win CE et al is pretty shaky IMO. Unless there is a point to an OS ISTM it is a waste of money and ROM.

Have look at

btw..

Do you need multi-threading? Memory management? Does the code require any kind of portability? How complicated is the system?

I'd be inclined to do the opposite. Even for systems of moderate complexity, C/C++ on top of a full-blown OS like Windows CE will allow uncomparable development speed.

-Le Chaud Lapin-

Depends. 100k+ units or $100k+ each?

100k quantity. There is no need for multithreading. Memory management isn't that complicated. In fact, the system needs only be quite simple, however it seems to be heading down the over engineered path. I am pushing to have the whole design, coding included to be outsourced.

The core reason of using CE is as previously stated, development effort. The other reason is so that the hoards of c# programmers can also do embedded devel on compact framework. For me the benefits are outweighed by the complexity.

Anyway, if anyone also has papers on OS v non OS relative costings I would also be interested in that. I am pissed that the Linux community has not fired back with a paper against the MS paper. There is plenty of attacking statements yet no one will provide figures.

This should definitely help:

--
Reply to nico@nctdevpuntnl (punt=.)
Bedrijven en winkels vindt U op www.adresboekje.nl

C#? Ufff! You're doomed. I cannot possibly convey via any digtial medium my digust with this language and the principles that it represents.

If you are in the company of people who insist on programming in C#, it's already hopeless.

-Le Chaud Lapin-

I actually dont mind C#. Its a bit of a struggle being a c++ person, but I am getting there.

Well that is where you are wrong. The core development done is aimed at visual applications for windows PC's. There is not a better language around IMHO more suited to this. Our embedded dev is very limited, so there is no need to use anything else.

I had a lecroy cro that spent all its time in the shop with front end noise problems. I have no confidence in there ability to design hardware, let alone software.

I spent today playing with CE6 and the compact .net framework. For an embedded OS MS have removed binary serialization and byte size packing of structs from the compact framework. The 2 things i want the most. The also support unicode by default, and not ansi. Poor implementation descisions i feel.

Pushing a decade back I had a 200MHz 4-channel mode of theirs that I liked (of the big box grey & blue models that had that silly "receipt printer" option); worked fine for me.

But they did come from more of a low(er)-speed, deep-memory, data logging/production testing environment than a "how fast of a vertical amplifier can I build?" kind of heritage like Tek did. Plus Walter LeCroy seemed to have a slightly over-inflated ego...

Perhaps (whooops, I just bought a second hand LA314H), but the last thing you want is a product that needs updates from Microsoft and/or can't be connected to a network together with other Windows machines.

--
Reply to nico@nctdevpuntnl (punt=.)
Bedrijven en winkels vindt U op www.adresboekje.nl

That depends. Programming GUI applications in C++ with plain .NET is kinda messy. The alternative is Delphi which gives nice GUIs with a crash-o-magic for free. So C# is an option.

For me the best thing currently around to get a GUI application done easely without code that sticks to a certain platform or vendor is wxWidgets.

--
Reply to nico@nctdevpuntnl (punt=.)
Bedrijven en winkels vindt U op www.adresboekje.nl

I still think Pascal is a better choice for embedded products.