| ??? 08/08/09 16:34 Read: times |
#168260 - Do you directly or indirectly work for Keil? Responding to: ???'s previous message |
Per Westermark said:
Richard said:
Do you seriously believe that those who'd steal a product if it were sufficiently difficult to get past the time-limiting features would ever buy the product? Yes, I seriously believes that a significant percentage of Keil's buying customers would be willing to use a cracked version for free, if they could. What do you consider a significant percentage? 2%? 10%? 25%? If the software is as easy to crack as would be likely with low-cost "cracked" version, the software can't be that good either, since the bulk of their effort goes into protection, rather than product quality. Now ... Which is "better", KEIL or IAR? What about the others? On what basis do you believe one is "better" than the others? How did you arrive at that conclusion? How many man-hours were dedicated to making that determination? Who paid for those man-hours? Was it Keil, or IAR, or one of their competitors? Why? Because they save $4000 by not paying a license. More than one company doesn't care about paying for used products or about support until they get stuck, and when they get stuck, they get angry that they don't get support, since they have a commercial (but unpaid-for) software.
If you where not thinking about the money - why would you then bother so much about thinking in investing so much time with the evaluation? I'm not so focused on the money itself, but on how I spend it. If one compiler reaches 90% of my requirements, and the other reaches only 85% but costs $400 instead of $4000, well, I need to know that, and I need to know which 15% of my requirements aren't being met by either of them. Since I will probably find that neither meets all of my requirements, then I need to know which of my requirements each one does meet. If the difference doesn't matter to me, then I'd be foolish to buy the more costly of the two, regardless of the price difference. After all, in some things, all other factors remaining equal, price is a reasonable measure of quality. Software is one of a very few notable exceptions. The tools does support mixing of assembler and C, so a normal developer would write whatever he can in C, and then add in whatever assembler the project requires. A customer who do want support would figure out that $4000 is expensive, but still a limited amount of money - every saved hour will recover back some of the money from the license. If many hours are involved in making the software comply with the printed documentation, or if the printed documentation doesn't cover every condition that can arise within the software, nobody should buy it. The support issue becomes moot, because this should have surfaced during evaluation. Of course, if the evaluation period/capabilities are sufficiently limited, blatant flaws are easily masked. That's what these software houses do, don't you know. They produce crap, dress it up, let the customers find the flaws and, eventually, fix some of them ... maybe ... Richard said:
Nobody said anything about "a time limit you can tamper with." If the software guys can't figure out how to prevent such tampering, they get what they deserve. Any timing scheme that isn't locked to a hw lock or requires cryptographic access to a server the user don't have access to can be broken quite easy. The software guys just can't protect themselves from disk mirrors, virtual machines, ... Too good attempts to protect yourself will result in a huge percentage of valid testers not being able to run the software. The only thing that works reasonably well is a tamper-proof device with non-volatile storage and encrypted or signed read/write operations. That still leaves the binary vulnerable to traditional cracking. And just so you know: PC programmers can be just as clever as embedded developers even if you are of the view that anything/anyone in relation to a PC is crap. A calendar-based eval package installable on only one machine would be quite straightforward to generate. It simply has to know the identity of the machine in question, BIOS checksum at installation * date * time, * NIC MAC address * registered evaluee's ID ... you get the picture. Hardware locks make it impossible to distribute evaluation versions by just having the prospective customer download the tools from the web. And any scheme that don't catches a company acquiring multiple locks will allow the company to just get a new hw lock whenever they need to release a new version. But wait a minute - your suggestion was not 30 days calendar time but 30 days usage time. So every hw lock the company can get their hands on (at way less than $4000) would represent a full 30 days of use. Many companies would probably manage with one evaluation license/year, giving them 30 days every year for updating the firmwares of their products ... ... and with an otherwise time-limited eval package, they simply freeze the clock/calendar on the machine on which they install it, and it "lives" forever, right? Registration with networked servers has the problem how you catch a customer that tries to register a second time when the first month is up. You don't need multiple computers to create multiple computer signatures. And it is hard to know what is a valid company in another country, so each employee could make a registration of their own.
Since people are cheap, a tool company must be prepared to ship quite a lot of evaluation versions for each sold one, so they can't afford much cost involved for each evaluation. In the end, a demo license that counts used days instead of calendar days can be very good for the user, but can be very dangerous for a company expected to ship few, but expensive, tools. It doesn't take many cheaters to get a 10% loss of sales. But such a model for evaluations isn't likely to add so many extra customers. Richard said:
I just used that as an example. The price isn't the issue. Of course the price is the issue. Why else are you so worried about just buying the Keil license? The tool adds something you do not have, but does not remove anything. You are still just as able to write your assembler code. Unless I was correct in a previous comment - you don't want to look too hard at the Keil compiler, because you may find it way better than you want it to be. Nope ... Price isn't the issue. Keil's Tech Support people blatantly and repeatedly lied to me about their product. As a result, I wouldn't buy anything from them, no matter what a value it was purported to provide. If they can't honestly and precisely tell me what, exactly, their product can, and, more importantly, can't do, and if their demoware won't allow me to determine that for myself, their product is of no interest to me. I had a full but time-limited version of IAR's product, but, not having the option of giving the IAR product my undivided attention at the time, its time limit expired before I could exercise it adequately. So far, it's been much easier to use ASM, specifically a macroassembler, with my library of macros, which I understand well enough to equal the comparable intrinsics of any compiler, and, as a result, I know the generated code is as fast and as dense as I require. That can't be guaranteed by any compiler. RE |
