Forums

STM32 ARM toolset advice?

Started by John Speth October 7, 2008


> CBFalconer wrote: > > > Chris has a simple and infallible mechanism. Open-source = bad. > > Expensive = good. > > P.J.Pauger has often said he can't afford open source. > > My direct experience with FOSS compilers > > 1) Optimization standards and metrics reflect several years old > technology. Depending on target a considerable difference. > > 2) The source level debug information generated in most FOSS compilers > > did not reflect the optimization choices the compiler made in code > generation > > 3) With a single exception none of the FOSS systems I have looked at > even attempted to incorporate silicon errata into the code generators > something we have been doing in our commercial compilers for a decade > and a half. > > You may actually get what you are paying for. > > Regards, > > -- > Walter Banks > Byte Craft Limited > http://www.bytecraft.com > > >
>Paul Black wrote: > >Chris has a simple and infallible mechanism. Open-source = bad. >Expensive = good. >
Open-source is not free, even if it is often suggested that it is. When I find a compiler issue I report it to the vender I pay maintenance to and within a few days their engineers have fixed the problem, and our projects are back on track. Since I am delivering commercial software to date and to budget that is essential, over runs of effort or timescale cost money – big money. But yes I pay a yearly maintenance contract for it. Yes it would be nice not to pay but I also pay insurance on my car, which I was glad of when someone drove into the side of my car recently – no hassle – no cost to me, just let the insurance company fix it, and have a curtsey car whiule mine is repaired. (I wonder do you insure your car or do you prefer a “free” alternative?) If on the other hand I have to rely on my engineers trawling the internet, and then diving into the innards of open source software to fix a compiler issue, that is not free, not remotely free. It actually costs me about £70 an hour in labour charges, (wages, national insurance, taxes, pensions etc). We write embedded software we are not compiler experts, even if one of our developers would love to play writing his own compiler . While the engineer is doing all that he is not writing the code our customers pay us for. So a single issue with the compiler could costs us many £1,000s and could even risk us losing customers and future business. If we lose a major customer because we are late while fixing problems in “free” open source and poorly supported tools people will lose their jobs. Now imagine that it is you who loses their job, because we decided to use a “free” open source tool instead of a fully supported and commercial tool. - I wonder would you thank me for using open source tools when you lost your job, would you be pleased you lost the job, would you still think open source is free, or would you tell everyone you met how unhappy you are that you lost your job and had to sell your house because your employer was a cheap skate and didn’t invest in commercial well supported tools. I recently abandoned the idea of using linux because I was horrified at how much using linux on a new project was going to cost. It was far cheaper to use WinCE. I am sure if I paid you the saving we made by not using open source software you would be delighted. In fact you could afford to buy a lot of “expensive” tools. Bocote
Bocote <PBlake@eden-electronics.co.uk> wrote:
>>Paul Black wrote: >> >>Chris has a simple and infallible mechanism. Open-source = bad. >>Expensive = good. > Open-source is not free, even if it is often suggested that it is.
Indeed. There is no shortage of vendors willing to sell you the support you want for most open-source products, including Linux and GCC. That way you get the comfort of having someone to yell at, *and* the advantage of being able to examine how things really work if you need to or want to. I don't understand why this is always seen as an either-or proposition when you can have both. -Anders
In message <48EFC70D.C9457295@yahoo.com>, CBFalconer 
<cbfalconer@yahoo.com> writes
>Paul Black wrote: >> Chris H wrote: >? >>>> Have you used any of them? >>> >>> Not personally. >>> >>>> Which was the best? >>> >>> No idea. >>> >>>> Do they have good debugging support? >>> >>> Not compared to commercial tools >> >> How on earth can you state that conclusion based on no experience >> and no idea? > >Chris has a simple and infallible mechanism. Open-source = bad
That I have found generally the case.
>Expensive = good.
No. However some open source is very expensive (see other email in this thread from Eden Some closed source is very good and FREE You seem to have the stupid idea that all closed source is expensive and all open source is free. -- \/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\ \/\/\/\/\ Chris Hills Staffs England /\/\/\/\/ \/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/
On 2008-10-11, Bocote <PBlake@Eden-Electronics.co.uk> wrote:

>>Chris has a simple and infallible mechanism. Open-source = bad. >>Expensive = good. > > Open-source is not free, even if it is often suggested that it > is. When I find a compiler issue I report it to the vender I > pay maintenance to and within a few days their engineers have > fixed the problem, and our projects are back on track.
The last time I reported a gnu toolchain bug, it was fixed within a few hours. :) I've also been ignored by commercial toolchain vendors when I reported a bug.
> Since I am delivering commercial software to date and to > budget that is essential, over runs of effort or timescale > cost money ??? big money. But yes I pay a yearly maintenance > contract for it.
You can get commercial support for open-source if you want. -- Grant Edwards grante Yow! INSIDE, I have the at same personality disorder visi.com as LUCY RICARDO!!
In message <gcqjt5$h1n$1@aioe.org>, 
Anders.Montonen@kapsi.spam.stop.fi.invalid writes
>Bocote <PBlake@eden-electronics.co.uk> wrote: >>>Paul Black wrote: >>> >>>Chris has a simple and infallible mechanism. Open-source = bad. >>>Expensive = good. >> Open-source is not free, even if it is often suggested that it is. > >Indeed. There is no shortage of vendors willing to sell you the support >you want for most open-source products, including Linux and GCC. That >way you get the comfort of having someone to yell at, *and* the >advantage of being able to examine how things really work if you need to >or want to. I don't understand why this is always seen as an either-or >proposition when you can have both. >-Anders
Having been doing tech support for may years I can tell you that giving most programers the source to fix a compiler is not a good idea. Most don't understand compilers. They also go on about "ANSI-C" !!! The other disadvantage is that it is old technology and is not as good at is primary purpose and the commercial tools. -- \/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\ \/\/\/\/\ Chris Hills Staffs England /\/\/\/\/ \/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/
In message <TcCdnUe6NbbdQm3VnZ2dnUVZ_rjinZ2d@posted.visi>, Grant Edwards 
<grante@visi.com> writes
>On 2008-10-11, Bocote <PBlake@Eden-Electronics.co.uk> wrote: > >>>Chris has a simple and infallible mechanism. Open-source = bad. >>>Expensive = good. >> >> Open-source is not free, even if it is often suggested that it >> is. When I find a compiler issue I report it to the vender I >> pay maintenance to and within a few days their engineers have >> fixed the problem, and our projects are back on track. > >The last time I reported a gnu toolchain bug, it was fixed >within a few hours. :) I've also been ignored by commercial >toolchain vendors when I reported a bug.
Perhaps it is your approach. :-)
> >> Since I am delivering commercial software to date and to >> budget that is essential, over runs of effort or timescale >> cost money ??? big money. But yes I pay a yearly maintenance >> contract for it. > >You can get commercial support for open-source if you want.
So they are not free.. Where is the advantage? -- \/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\ \/\/\/\/\ Chris Hills Staffs England /\/\/\/\/ \/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/
<Anders.Montonen@kapsi.spam.stop.fi.invalid> wrote in message 
news:gcqjt5$h1n$1@aioe.org...
> Bocote <PBlake@eden-electronics.co.uk> wrote: >>>Paul Black wrote: >>> >>>Chris has a simple and infallible mechanism. Open-source = bad. >>>Expensive = good. >> Open-source is not free, even if it is often suggested that it is. > > Indeed. There is no shortage of vendors willing to sell you the support > you want for most open-source products, including Linux and GCC. That > way you get the comfort of having someone to yell at, *and* the > advantage of being able to examine how things really work if you need to > or want to. I don't understand why this is always seen as an either-or > proposition when you can have both. > > -Anders
This is just the same old arguments that have been put pro and anti open source for years. Some open source code is good and well supported, some bad. Some closed source code is good and well supported, some bad. Etc, etc. Some open source comes with commercial support, some doesn't. Blah, blah, blah. As with everything in life generalisations are usually misguided. I wonder how many anti-open source people use Firefox? -- Regards, Richard. + http://www.FreeRTOS.org & http://www.FreeRTOS.org/shop 17 official architecture ports, more than 6000 downloads per month. + http://www.SafeRTOS.com Certified by T&#2013265948;V as meeting the requirements for safety related systems.
Chris H <chris@phaedsys.org> wrote:
> Having been doing tech support for may years I can tell you that giving > most programers the source to fix a compiler is not a good idea.
Usually being able to look and understand the implementation is enough. For instance, working on OS X having access to the IOKit sources has saved my bacon several times - either by clarifying vague or incomplete documentation, or by being able to show that some particularly tricky bug was in the operating system and helping us implement a workaround until the vendor released a fix.
> The other disadvantage is that it is old technology and is not as good > at is primary purpose and the commercial tools.
There you go again, confusing "commercial" and "open-source". Though I suspect in your case it's intentional. -Anders
Chris H wrote:
> In message <TcCdnUe6NbbdQm3VnZ2dnUVZ_rjinZ2d@posted.visi>, Grant Edwards >> On 2008-10-11, Bocote <PBlake@Eden-Electronics.co.uk> wrote: >>> Since I am delivering commercial software to date and to >>> budget that is essential, over runs of effort or timescale >>> cost money ??? big money. But yes I pay a yearly maintenance >>> contract for it. >> >> You can get commercial support for open-source if you want. > > So they are not free.. Where is the advantage?
The advantage is (and this is not at all a new argument): if the commercial vendor tells you "I won't fix this bug, buy our upgrade instead", "You are using a non-standard product which is not covered by our support contract", or "We won't fix this problem, implement your own workaround", that's it. Maybe you can go to court and solve the issue, so you get a fix six months later. With an open-source tool-chain, you can always hire someone to do it for you (if you don't want to do it yourself). The open-source people will also not ship broken dongle drivers that crash your development machine, nor will they tell you "of course we delivered you the CDs and dongles, but we will not give you the license files until you sign this contract expansion which limits your rights". None of the above has been made up, everything experienced during my (still short) life as embedded developer. I would prefer open-source by far. Actually, aside from the Windows and the toolchain on my development computer, all programs I use regularly are open-source. Unfortunately, the GCC for our target is pretty bad. Maybe my experience also comes from the fact that we're a medium-sized European company, which doesn't have high priority for glorious American companies. If we were AT&T or the DoD in the United States, our live might be easier. That aside, our company policy is to buy source licenses for everything where we can get them. This has saved our lives, nay, projects, more than once in a while. If there were a source license for the compiler, we'd probably get it. Compulsory car comparison: would you buy a car where the motor block is cast in concrete, which only the vendor can fix, and only if you buy a support contract? Or would you buy one which every backyard mechanic (or you yourself using a self-help book) can fix? Stefan