The Common Linux Misconception [Archive]

Andy Sjostrom

07-10-2003, 09:55 PM

<div class='os_post_top_link'><a href='http://palmtops.about.com/cs/alternativepdas/fr/Yopy3700.htm' target='_blank'>http://palmtops.about.com/cs/altern...fr/Yopy3700.htm</a><br /><br /></div>About.com has published a review of a Linux PDA called <a href="http://www.yopy.com/english/products/yp3700_overview.htm">Yopy 3700</a>. A number of things comes to mind when reading the article! I agree with the article author that the Yopy 3700 has a unique form factor and while it might not fit my needs, I do believe that such a form factor has its raison d'être. The device does not otherwise stand out from a hardware feature perspective, maybe with the exception of the 128 MB RAM. I would love to see a Pocket PC looking like this.<br /><br />What really caught my attention, however, was to again see the misconception that Linux in commercial mobile devices is free. This is a quote from the article: "While the cost of licensing the Palm and Pocket PC Operating Systems can be prohibitively high, Linux is based on the Open Source platform, eliminating those high licensing costs. Linux is a great opportunity for a start up to compete with the big boys."<br /><br />This line of reasoning does not make sense. First of all, a Pocket PC license is less than $20. The Yopy costs $499 meaning that a Pocket PC license cost would mean no more than 4% of that cost. Those 4% would buy an OEM a common and tested platform, access to millions of developers, thousands of applications and so on. To me, that "is a great opportunity for a start up to compete with the big boys".<br /><br />Instead, from an OEM perspective, when choosing Linux to put in your mobile device you either a) license from someone who has put together a working Linux distribution along with the core PDA applications, or b) build it yourself which brings us to the classic build vs buy decision. I doubt that anyone can put together a mobile device operating system, applications, create a development platform, development tools, work through internationalization, testing, that matches the Pocket PC to a cost lower than $20 or less than 4% of the price of the device. While Linux might have price advantages on the desktop, it certainly does not in mobile devices.

darrylb

07-10-2003, 11:05 PM

I have one word for Linux solutions that are technically good but stupid from a business perspective: ZEALOTWARE

This thing is zealotware - their business thinking is apparently limited to "we must do linux". The things you present above are valid Andy, and I agree wholeheartedly with you. Many linux zealots are completely devoid of business sense and probably dont know what a business case is...

Aside from the fact that I think this thing looks like a scaled up StarTAC mobile phone (which I also hated), I have to ask why? 8O What compelling features does this have that a product by a bigger company does not have? A bigger company that is more likely to be around in five years? A bigger company that can provide support for my stupid consumer questions?

I really dont see this going well.... :roll:

Hooked

07-10-2003, 11:33 PM

I agree, Linux is not a free lunch as it is often portrayed.

But, for a device manufacturer there is probably more flexibility licensing a Linux distribution in some ways compared to PocketPC and PalmOS.

No required hardware specs (i.e - PocketPC) and with Linux you get the source code, which you can modify as you need.

Linux distributions have fewer political implications compared to either Palm or PocketPC, and are less able to "punish" licensees for not "towing the company line."

Linux is still a bit immature for most consumer devices, but that's changing (see http://www.eetuk.com/bus/news/OEG20030701S0010)

PalmOS and PocketPC are clearly the obvious choice for the generalized, commodity market, because the applications and consumers are already there.

But for products which have a specialized purpose or niche and taking advantage of those pre-existing applications is not a priority, I do not see a clear advantage to choosing Palm or PocketPC.

In the case of this device however, I do not see a compelling reason to choose Linux, unless using Linux is the specific need. I do like the idea of a folder-type form factor. But, from the specs, this thing seems like a step back from the Zaurus offerings. I'd rather have the Sharp SL-C760, although not at $800.

I do wonder why Linux is not used on mobile phones.

jmarkevich

07-10-2003, 11:44 PM

sfjlittel

07-10-2003, 11:59 PM

I am subscribed to a mailinglist, and a few days ago I read that the gui I like (opie) is fully translated in my language (dutch) by someone on that list. I have not seen Windows Mobile in my language...

droppedd

07-11-2003, 12:59 AM

Don't confuse free "gratuit" with what should be free "libre". PocketPC seems to be somewhat inflexible to OEM's needs -- the best they can offer is add-on applications, and maybe drivers.

What if someone wanted to rewrite the tasking scheduler? Make a completely different input model than SIP? How about HI-RES screens that rotate? Blow memory limitations off the map?

The OEM can do this. Actually, the USER can do this, since GPL requires the code be included. This kind of flexibility is what embedded devices really need, and Palmtops don't desperately NEED but can be helpful.

ok... I agree that the ability to mess with things like resolution or hardware architecture can be fun, and even useful in the small picture, and I think it's wonderful for users to be able to mess with their OS more than PPC lets them (not even a builtin regedit!). However, as much as I am "pro-linux" I think that allowing those things you said to be up to individual manufacturers can be a very bad thing. Case in point with Palm. There was a point not too long ago that between the standard resolution 160x160 coexisted with the Sony Clie 320x320 along with whatever weird res the soft-graffiti HandEras had. This was a royal mess for developers, as you had to code using everyones APIs and such. It's an awful place to be in; so in that sense the standardization of PPC hardware and software by MS is a good thing. It's less hassle for everyone, and less confusing for endusers.

That said, non-standard Linux has a place. Mostly, I think, for corporations that need a very customized system. For that, on both servers and on PDAs etc. Linux has what it takes as long as your primary issue is not interoperability with existing software or cross-compatibility with new devices. But for PPCs? what a nightmare. I can't even get half the software I want working on the new 2210- imagine if it also had a diff. resolution, a different processor core, even a completely different method of entering appointments. That would be a royal mess.

That said, good point distinguishing between "free beer" and "free speech," as it were. Interesting that some languages have proper different words for them.

Perhaps the FSF should change its name from "free software foundation" to the "Freedom Software Foundation" to clear that up (hey, they get to keep their initials, too!). It has the added benefit of sounding patriotically American (although people may think it's the new name for the French Software Foundation... Freedom Fries, anyone? :wink:)

leshman

07-11-2003, 01:23 AM

What about development costs? I was under the assumption that MS was not going to release any more versions of eVC++ or eVB. What is the cost of Visual Studio .Net Pro? About 500 bucks. And besides the license cost, what about the cost of creating your customized version of the OS? Here's a quote on the cost of platform builder:

Platform Builder costs $995 per developer through March 7. After that, it will sell for $2,995. A fully functional version is available for free at www.microsoft.com/windows/embedded/ce and will last 120 days. The software can target ARM, MIPS, SuperH and x86 designs.

So now you are talking about $3500 (Platform Builder and Visual Studio) per programmer to create your OEM version of the plus. That's a little more than just $20. And no, I am not a Linux user. I just want people to have all the facts before they go on a rant.

szamot

07-11-2003, 01:38 AM

Articles like this perpetuate negative stereotypes about MS and promote ignorance by falsely enticing users to believe that if everyone was using Linux the PDA's would cost $1.99 or better yet be free. Shame on you for people for complaining about licensing cost, if you don’t' know the facts investigate them before you make a fool of yourself by writing an article like this. Let’s be hones about this Linux mystery on PDA’s, this unit has some nice features like memory, battery and the design but:
• “Linux can be intense and confusing to a new user
• Hardware compatibility
• Lack of Syncing capabilities with Windows and Mac
• And probably the biggest drawback is the Lack of Software compared to Palm and Pocket PC”
….Which means that all this PDA is good for is for duck taping it to the front of the car and pretending you are Night Rider. But the flaming chicken Trans AM wasn’t cool then and it certainly is still not cool today.

darrylb

07-11-2003, 01:59 AM

What if someone wanted to rewrite the tasking scheduler? Make a completely different input model than SIP? How about HI-RES screens that rotate? Blow memory limitations off the map?

Like WisBar, Gigabar, etc. For a SIP, you mean like Fitaly or transcriber (how many more ways do you need to input text). With HiRes Screens, are you meaning like with VGA output or do you have a magical way of increasing the static resolution of tft screens that I have not heard of? Memory limitations? There is only so much 64 Mb RAM can do.

The OEM can do this. Actually, the USER can do this, since GPL requires the code be included. This kind of flexibility is what embedded devices really need, and Palmtops don't desperately NEED but can be helpful.

OK, so if you want extra functionality, you will recode it? Great. At least with PPC you have the option of C/C++, C#, VB, etc, with linux you have to know C/C++ - What percentage of end users know C++? < 1%? So how is this helpful to the user?

My point is that the wonderful benefits that linux offers come at a price, and why should the end user have to do all this stuff when the functionality is already there for Pocket PCs (and probably PalmOS)? Making Linux based PDAs does not make sense at this stage unless you are a big company that can afford to do lots of extra work (ala Sharp) to ensure a high number of useful devices.

If I was an investor, I would be extremely cagey about investing in a company that was trying to do a Linux PDA.

konfoo

07-11-2003, 02:25 AM

Bringing up the cost of platform builder is a moot point. If you make or break business decisions for a PDA line based on a $3000 investment then you should get out of the business. $3000 is a drop in the tank.

There is confusion about 'Linux being free' and 'Linux development being free'. Development time and cost of resources is where the real cost is. And you are not going to get Linux developers to code your base platform and do integration with hardware gratis.

As for cost of Linux for embedded platforms, its not free either. Usually you source a base system from a commercial entity selling a customized Linux distro for embedded use. This costs $$$.

Anyone who complains they can hack such a system together in their backyard since all the GNU tools are free are living in a dream world and should stick to slashdot.

As for all the arguments about Linux being confusing - moot point! Its all up to the interface you put on it.

Linux isn't used on mobile phones because the big few (Nokia, etc.) get revenue from selling their platform development kits. Vendor lock-in = profits.

felixdd

07-11-2003, 02:45 AM

Call me a general consumer but...what's the advantage of having Linux on a PDA form factor?

gorkon280

07-11-2003, 03:52 AM

with linux you have to know C/C++

OR PERL, bash, fortran,cobol....etc.etc. THAT is the beauty of Linux. I realize there is development issues mostly on Linux. But the thing is is even after Sharp concedes and decides not to make Zaurii, there still will be Open Zaurus. There STILL will be an option for users who chose the Zaurus. This is unlike those Pocket PC's who used non flash roms and also the ones who have discontinued support (do you need to ask if Casio's e200 is getting a WM 2003 update?). I have actually picked up a Zaurus and checked it out and it seems quite useable. Linux FLEW on the Zaurus and it had Jeode on it as well. Hardware support, strangely enough, is MUCH wider in Linux. If it worked on other platforms, it should work or be easier to get to work then Pocket PC. Folks can also write drivers for these items. Who writes them for Pocket PC? Either the company who made the device or Microsoft. In the case of another article posted today about the CF problem....this could be fixed in a day or so by Linux hackers. Also, since it is Linux, in some cases, depending on what a program does, you can usually recompile something and it works on the Zaurus. Granted, some useres won't want to but as long as there's tools for packaging applications ( and there are ) it only takes one person to compile it and post it. With a Zaurus, it's even possible to take some devices such as that SD Digicam that is only for palm and write apps and drivers to make it work on a Zaurus. Granted, not every user will want to do any of this but like I said as long as someone can package it and make it easier to install anyone can install it. So, I guess I can kind of see a reason to run Linux on a PDA. My reason is a little harder to get over. How many open source mapping programs are there? None! :D

szamot

07-11-2003, 03:55 AM

szamot

07-11-2003, 04:03 AM

aramp

07-11-2003, 05:44 AM

One could and should be biased when talking about personal opinions. But when talking numbers, one should be clear-minded.

Andy says PocketPC is USD20 for an OEM. Hopefully it's so. Anyway, it is not 4% at all. When the price for PPC is somewhere around USD300, it's street price. The OEM gets at most 50% of it. Actually with all rebounds and channel discounts, they may well end up with less than 40%. Meaning what OEM gets for the hardware it lets out is some USD150. Now let's calculate his profit, not the revenew, which is of no interest to us. Color TFT screens are rather expensive and cost not less than USD30-40. Add memory ($20), CPU($10), other components ($30) production cost ($10), etc, and you will find out that the OEM's margin on each device is at most USD30-50. Add advertizing costs, add first year gurantee costs, they will eat up another USD20 from each device at least. Then USD20 to be paid to MS is HUGE, it's hurting, it makes the whole game almost worhtless to play unless you sell very large amounts. And this is the point. When you sell large amounts (I mean hundreds of thousands at least, actually millions), you have to pay to MS millions and tens of millions of dollars. Will it make sense then to spend a million or two to make your own OS and stop paying almost all of your profit to Microsoft? Sure it is reasonable.

The only problem is that consumer needs trademarks. When you are CASIO you don't need trademarks: make your own OS, sell whatever you like through a well-made dedicated channel along with watches, calculators, etc and be happy. But when you are, say, SHARP without such a dedicated channel, you have to build something that channel will accept. Linux is mainstream at the moment, with so many companies behind investing into marketing hype (including IBM, Intel), and it would be stupid not to make use of that. And there is an obvious audience they are targeting: people that like to think differently (the same strategy used by Apple). Both Palm and PocketPC are in now way striking anymore, they are commonplace. They are so many people around having them. So people that whant to look different and amaze others will like Linux device. Normally that's about 10% of the potential market. Not bad.

So the whole strategy is pretty obvious and I would really be surprised if it would be otherwise. Because normally companies have well-paid marketing teams doing surverys and looking for chances on the market. There is a technical challenge however, and one has to meet it. As far as I know, Sharp did well enough, I am not sure about this Yopy thing. But when you want to make a competitive product that no other may easily beat, you have to face challenges.

sfjlittel

07-11-2003, 09:55 AM

I know that Compaq at one point in time supported Linux development on the iPAQ, I should know they fixed my sick 3850 when it got a nasty case of bad Linux ROM, but have not heard anything new on it. There might be a place and time for Linux on the PDA's but not today and not any time soon. It needs functionality, support and lots of working apps to be considered ready for the mass market.

Compaq (HP nowadays) still supports linux on the ipaq. Take a look at www.handhelds.org. If you want to know what's going on then I suggest you subscribe to a mailinglist like "familiar" or "gpe" or "opie" or "ipaq". You can find the links to the pages where you can subscribe on the page I mentioned earlier.

aroma

07-11-2003, 12:47 PM

Just wanted to add my 2 cents...

First, as some have already stated, I would think one of the biggest reasons an OEM would go with Linux over MS/Palm would be the licensing hoopla / restrings the main two would require.

Second, as far as cost, sure $20 per deivce might seem like much, but if I were planning on manufacture say, 100,000 of these devices, that $2,000,000. If you can license the Linux distro for say $5, then thats a savings of 1.5 million dollars. Profit to put in the pocket, or piss into the wind... thats no small change if you are a smaller OEM.

- Aaron

jmarkevich

07-11-2003, 01:45 PM

What if someone wanted to rewrite the tasking scheduler? Make a completely different input model than SIP? How about HI-RES screens that rotate? Blow memory limitations off the map?

Like WisBar, Gigabar, etc.

Not a task SWITCHER, a scheduler. I'm talking kernel level and you're talking user level.

For a SIP, you mean like Fitaly or transcriber (how many more ways do you need to input text).

No, those all use SIP. I'm trying to illustrate the next big thing might come along (tilt sensor for input, Nintendo PowerGlove, who knows?) and Linux can respond faster to big changes like that.

With HiRes Screens, are you meaning like with VGA output or do you have a magical way of increasing the static resolution of tft screens that I have not heard of? Memory limitations? There is only so much 64 Mb RAM can do.

Again, we're on different pages. I mean OEM, you mean user. Say HP wants to put 640x640 screens in their iPaq. What are they gonna do? I mean say SanDisk comes out with 64GB expansion cards. What happens now?

The OEM can do this. Actually, the USER can do this, since GPL requires the code be included. This kind of flexibility is what embedded devices really need, and Palmtops don't desperately NEED but can be helpful.

OK, so if you want extra functionality, you will recode it? Great. At least with PPC you have the option of C/C++, C#, VB, etc, with linux you have to know C/C++ - What percentage of end users know C++? < 1%? So how is this helpful to the user?

If I'm an OEM, yes, I will code it. I will rewrite the whole kernel if I want. Probably in C/C++. Remember I never said end users HAVE to do it, it's mostly that end users can benefit a lot from what the hackers do.

My point is that the wonderful benefits that linux offers come at a price,

And my point is, same story with Pocket PC, and sometimes that price is virtually impossible to meet.

Andy Sjostrom

07-11-2003, 02:42 PM

I do wonder why Linux is not used on mobile phones.

Motorola is on their way with Linux phones. Look out around Christmas...

Andy Sjostrom

07-11-2003, 02:46 PM

Andy Sjostrom

07-11-2003, 03:03 PM

aramp! Welcome as a registered user! :D Your comments and thoughts are very well put. I agree with most of what you say.
The mathematics of my post are quite easy to agree on. First, the Pocket PC license is less than $20. How much less depends on each OEMs licensing agreement. But since $20 is an easy figure to use we'll stick to it.
$20 is 4% of $500. So, from a consumer perspective, $20 of my money goes to Microsoft. From an OEM perspective it is a different story. Assume they get 50% ($250). Then $20 is 8%. I don't think that is unreasonable that the actual software that the user uses costs as much as, for example the memory.

I also don't think "one million or two" takes anyone anywhere in terms putting together a solid, commercial operating system for mobile devices. Add to that the costs involved in staying on top of all the rapid changes in this market...

I think your analysis why people would choose Linux makes sense and your 10% assumption does not sound unreasonable either.

Thanks for contributing!

gorkon280

07-11-2003, 03:37 PM

There is still support for 8 - Track but so what. Linux needs to pick its battles, I don't think this is one of them, at least not yet.

Why does it have to pick it's battles? Linux is not driven entirely by a corporate entity. It runs on nearly everything. It does not hurt Linux developers to develop a cool new app. It does not hurt folks who are porting the kernel to other processors. They are doing their work for free lots of times. What monetary cost is it costing if they want to do it? None. Companies who build the handhelds reap the benefits.

I do realize that some software just is not available for it (like mapping software.....only one that comes to mind but I am sure there's more). Now where I DO see Linux surviving in a handheld format more is in a situatuion where you must develop a custom app and use custom hardware to get the specific situation done. This case it's perfect for it and the cost of developing the software would be much cheaper. Also, just because the current product does not give grandma warm and cozies(face it...grandma still loves paper) does not mean it's useless or not relavant. Personally, I LIKE having Linux in the fray competing. Competition will drive innovation on all of the platforms. Just look at Palm. For a while, they were king of the heap. Palm OS stagnated. Microsoft came out with Pocket PC and then licensed it to Compaq....LOOK OUT! There came a juggernaut that forced Palm to not only improve the OS but improve the hardware (ie, no more dragonball processors). Granted, the 20 dollar licensing fee per device is not alot, but companies who develop Linux handhelds are not held back by Microsoft requirements. They can include any piece of hardware they want with NO corporate entity saying no you can't do that. Pocket PC has some specs you can't exceed like screen size (STILL 240 x 320) and Flashrom Being needed (although this is a good thing). Linux can do the same for Pocket PC. Also, when was the last time you actually picked a Zaurus up? ALmost everything you can do with a PPC can be accomplished on a Zaurus. It DOES sync with Windows. It's sync may not be as elegant as Hotsync, but how can it be any worse than Activestink!

To go on record, I LOVE my Pocket PC and even I agree that, for me, Linux on a handheld computer does not make sense. Add Mapping software and this changes ;). Even then, the hardware would have to have a hair more to appeal to me. How about built in wired ethernet as well as WiFi?:) I love my PPC, but I do not dismiss other platforms right away because you are looking through PPC colored glasses. Try to be objective about it before you dismiss it entirely.

gorkon280

07-11-2003, 03:58 PM

If you can license the Linux distro for say $5,

Try free. Except maybe SCO and as far as I know, they have not sold there embedded stuff to anyone yet for a handheld (SCO isn't in too good of shape with managers now thanks to their frivilous lawsuit against IBM) that is actually selling Linux for embedded devices (which would included handhelds). Even if there was (say Redhat), the only thing that would be restricted from use are non GPL'd parts of Red Hats codebase and there is not much in RedHat that is not GPL'd (some installation scripts and code and maybe a few other things like RHN). Also, RedHat does not really care if you buy Red Hat and install it on 500 machines. So, the OS cost for a 500 installations would be 50 bucks. Even then, you can always grab all of the GPL'd code you want and recompile, manipulate all you want (as long as you give the code back) to make it run on a handheld. ANYWAY, (sorry to go off on a tangent) if going with Linux saves you 2 million on a product line, you'd go with it. In the big picture of things, for a upstart, Linux is perfect for handheld use. I would rather them use that 20 bucks to develop apps to run on it or adding hardware(like a REAL microphone input). In any case, Linux can be viable on it and it should be worked on as long as folks are interested in it.

I also would like to point out that one of the most successful home av components out there now is the Tivo. It runs on Linux and grandma's use it. Linux CAN be made handheld friendly and can be made useful by mere mortals.

aroma

07-11-2003, 04:14 PM

If you can license the Linux distro for say $5,

Try free.

I still don't think your going to get it for free. If I (as an OEM) am going to use a commercially viable linux build on a handheld, I'm going to have two choices.
(a) Do the work my self and develop my own distro for the handheld.
(b) License someone elses work, such as QTopia.

If I go to someone who has done the work on a handheld distro and say and tell them I want to use it on 100,000 new devices, they're not going to let me get away with licensing one copy of their distro and then using that for all my devices.

- Aaron

droppedd

07-11-2003, 04:22 PM

I also would like to point out that one of the most successful home av components out there now is the Tivo. It runs on Linux and grandma's use it. Linux CAN be made handheld friendly and can be made useful by mere mortals.

Tivo=/=handheld. Tivo is ONE product from one manufacturer that has little or no need to be able to install software or interact with other devices. Handhelds are another story; standardization is the lifeblood of the things, on a commercial level (as much as i love tweaking within those bounds with custom software).

That said, Linux embedded on some devices can be MUCH more stable than comparable windows systems. Example: the big old screen in Herald Square, NY above Macy's was down the other day with a windows Virtual Memory Error window on the screen 8O (i had a photo, but it got messed up on the stupid memory card... grrr.). But I think without a fully standardized version of linux for PDA (which would be counter to what the Linux folks here seem to think the benefits of it are) it's no good for PDAs. I use linux very often; it's the standard in my school for computer science. I just don't want it on my PDA anytime soon.

jmarkevich

07-11-2003, 05:59 PM

Don't confuse free "gratuit" with what should be free "libre".

I don't. This article compares Pocket PC with Linux and says that using Linux would eliminate license costs. I say: if that's true, then you'll have to add investments costs around development, testing etc. Otherwise, you will need to license someone else's work.

Uhh. Then remove all reference to the word "cost". Don't worry, that's a subtle trap. Just pretend Linux is $50 and you can do whatever with it. More like buyng a company than buying a product.

Yes it does remove licensing costs. If you want to make a product, you ALWAYS have some measure of development and testing. You could also Pay the US$20 to MS for PPC, develop and test your value-added software without insight to the underside of the kernel (fake it with published APIs), then gripe and moan for YEARS AND YEARS about something Pocket Word.

Why does Pocket Word not have inter-paragraph spacing? Why does it keep changing back to the default font? Can I write a module to get it to read OpenOffice? If I had the code, I might be able to figure it out. But that freedom has been withheld from me.

aramp

07-12-2003, 08:30 PM

aramp! Welcome as a registered user! :D Your comments and thoughts are very well put. I agree with most of what you say.

Andy, thanks for a warm welcome. Actually I was reading the PocketPCThoughts for more than two years, but this is the first time I decided to take part. :)

I also don't think "one million or two" takes anyone anywhere in terms putting together a solid, commercial operating system for mobile devices. Add to that the costs involved in staying on top of all the rapid changes in this market...

Right, but do not forget where these things are coming from. It's not Europe where the average programmer costs at least 100 000 EUR per year. It's not from US. It's from Asia. People there get moderate salaries. So $1-2 mln is a good chunk of money to do something big there and may stand for average of 30-40 man-years. Not bad, yeh... I know that from my own business experience. When it comes to US, I don't think it costs less for HP to adapt PocketPC OS received from MS to iPaq.

But the major question remaining is quality. Quality software development is still a challenge in most of the regions of the world. And when MS with its huge investments in QA and total quality management makes those annoying mistakes that become our nightmares, I can only guess what trouble are these Linux gadgets.