Apple

It is hilarious to read how Apple failed to sell me

1. in 2007
2. in 2008
3. and in 2009

I do love the company, I just find the premium too high for me personally but worth every penny for most. It would be fun to list what Apple did manage to sell me...

1. B&W iPods from around 2005 that I still use today. Well, one died around 2010 (battery did, hdd got replaced with CF a lot earlier), but the one I bought for Lili is still going despite hdd getting damaged by Katherine in 2010. Ebayed in 05-07 at $80 vs. $200 MSRP.
2. 2008. iPhone - the original, aka 2G. Bought in May 2008 for $250 from AT&T refurbished (2 weeks old upon arrival) and without contract (Thanks, but no $400 iPhones on 2 year ATT in 2007 and no iPhones or any other phones with contract for me since 2004). Was bought for iOS development. Still used today mostly as an MP3 player and for development too. Still last a week on a charge or 4 hours as a bluetooth MP3 player.
3. 2010. iPad, bought in May 2010 at MSRP (after a week waiting list) to give out to my sister. If it was not for the occasion would I buy it? Probably YES, even though I am happy that I did get a chance to buy iPad2 instead. However I would probably end up buying Nook Color for $200 instead of $500 for iPad2 that is worth every penny.
4. 2010. 3!!! iPhones 3G - all as gifts to family in Russia, ebayed between $160 and $200. Enough said.
5. 2010. ATV2 - bought the next day it become available, returned after a week since Boris chewed remote and replaced a week later with another one. At $99 it is the best Apple product yet once the community ported XBMC onto it and no longer do I have to have a Mac running to watch TV.

I am also ready to buy IPT4, but Katherine is dragging... So overall it looks like Apple is doing a lot better now (should I say Apple did great in 2010?) and I am buying their products once every couple years and keep on using them for 4+ years which is a good track record lately, but would hardly match a Dell that I used 1999-2009 or a Ford that I bought in 2000 and still use, or 1997 BMW that I bought in 2005 and continue using...

P.S. I did use non-unibody MBP (2007-2009) that I liked and 2010 aluminum i5 based MBP that I liked less at work, if that matters or counts...

Sysadmining or "a post I didn't finish in over a year" or what used to be "4 boxes 4 kernels"

Hazard installs 10.6.0 in about 15 minutes screwing up existing bootloader (fixable by repairing with Windows disk) - select just what is really needed to boot just once to apply combo update. Still the best way to install OSX is to CarbonCopy reference disk (with combo update applied) and to install Chameleon, set up Extra, etc on per board basis. Which is as follows:
GA-H55M-S2H needs

1. ApplePS2 driver deleted in /System/Library/Extensions for PS2 keyboard to work with VoodooPS2
2. IOUSBFamily.kext rolled back using Multibeast for 10.6.6
3. RealtekR1000L.kext added to /System/Library/Extensions to enable Ethernet
4. Chameleon RC5
5. My Extra folder with DSDT.aml mapping RTL888b paths and the rest of extensions to enable PS2 and audio and "busratio=20 arch=x86_32" in Boot.plist

AMD

HP board still doesn't work with PS2 keyboard no matter what I try (don't bother installing VoodooPS2 from Multibeast because it would kernel panic on AMD and SL_Voodoo_PS2 didn't work either). Also, AMD is unstable in 64-bit mode (with -force64 option that is - Finder crashes, quicklook doesn't work, etc), so especially for OSX AMD suck even more. Just run in 32-bit on AMD.

Still, AMD requires legacy_kernel (installable as package after combo update) and HP mobo requires FakeSMS (in Extra - delete the one installed by Hazard in SLE), EvoReboot and ElliotForceLegacyRTC not to fuck up BIOS, optional NullCPUPowerManagement, plus VoodooHDA in SLE - all installable with Multibeast.

GA-G31MX-S2

GA-G31MX-S2 doesn't boot with Q6600 anymore. When it did, new Hazard (Intel only) worked fine for 10.6.2 that is. The following is too old to matter but anyway...

• Chameleon_RC4
• CMOS_Reset\Elliot,
• Patches\About this Mac
• Patches\evOreboot
• Patches\PS2 Fix
• Audio\Azalia and ALC888
• nVidia\NVEnabler and NVInject

EEE

Almost 2.5 years and EEE 1000H is going strong (stronger then the rest of $300 computers I had since 2006)... in its class that is. EEE 1000H is the reason I want another $300 EEE or any netbook to still last around 4 hrs after 2.5 years of battery abuse. If it was not for 1024x600 graphics, shift key and failing trackpad it would have been perfect despite that you cannot do any Android development on it and to do WP7 you need 2GB RAM and iOS development ain't fast. Still EEE would be fun when Lion comes around or Windows 8 for that matter.

Atom needs patched kernel, but some managed with stock on H1000 (hmm, really?). Apart from that just use backed up Extra.

Windows scores

EEE	2.2,4.5,2.0,3.0,5.4	GMA950 is the lowest and processor is not fast either
D605 2.67GHz	4.3,3.9,1.0,1.0,5.9	ATI-200 is shittier then GMA950
HP	5.6,5.5,3.8,5.2,5.7	GF8400 lowest score is "aero" with "gaming" score higher then HD3650
GA-G31MX-S2	5.9,5.5,4.9,4.7,5.9	HD3650 lowest score is "gaming"
GA-H55M-S2H	6.8,7.2,4.8,5.1,5.9	i3-530 integrated graphics is on par with HD3650

Sizing disks

No matter what, OSX needs at least 40GB... well one could squeeze it into 30, but that's really pushing...

OSX takes about 7.2GB. Add to that 10+GB for XCode and on 30GB drive we have just 12GB left. FCP would take another 1.5GB making it impossible to install iOS SDK updates. Premiere would take from 1.5GB to 2.4GB alone leaving just 8GB available and making 30GB OSX drive hard to update.... So, for OSX allocate 40GB. CS5 is just a 64-bit update of CS4 and since I have no NeatVideo for CS5 for now we are editing in CS4 and thus 32-bit CS4 is what goes onto OSX.

Win 7 takes over 10.5GB after install. Turning Games, Tablet and Gadgets off and running update would bump it up to 11GB. IE9 would make it 11.5GB. Once you install Office it is 13.5GB and after update we are talking 15GB system. Premiere 1.2G-1.4G plus update and we are past 15GB and 20GB is too small. Thus Win 7 needs 30GB for system drive.

40BG+30GB makes 70GB and that's just for software alone. I guess I am not buying $999 11" MacBookAir, not until it gets 128GB SSD (screw just 5hrs battery too). At $1199 13" MacBook Pro is twice the weight of 128GB MacBookAir and twice the value and lasts "7 hrs" too. But then again, why would I buy 2 lbs, 2.3GHz i5, 1280x800 GMA based system for $1200 when since 2006 I was buying $300 computers?

P.S. I do detest the time wasted to learn the above truths.

HTC Touch Pro on Boost

1. HTC Touch Pro on ebay for around $150
2. Convert Boostmobile iDen account to CDMA
3. Profit

Profit includes EV-DO Rev.A data for $0.35/day that open up lotsa possibilities.

• Tethering from WinMo
• VoIP from WinMo
• Flashing with 6.5 ROM (MightyROM6)
• Installing Android

Following are general links for HTC Touch Pro

• Howards Boost CDMA
• PPC Geeks instructions to flash CDMA ROM
• xda-developers HTC Touch Pro CDMA thread

It is not clear if Haret booted Android could be used to debug Android apps from Eclipse.

Of all the ROMs for HTC Touch Pro out there Herman is the best (SKI and Freedom quite suck IMHO). I like MaxSense the best, second best would be Cookie, Titanium is plain WinMo 6.5. All Herman ROMs cannot provision for Sprint data, so anytime I swap ESN, I need to flash stock Sprint ROM.

WinMo 6.x

Stock WinMo 6.x is truly pathetic. The best that MS could show when it comes for touch enabled devices is a hack-hook into WndProc to sometimes recognize gestures on WinMo 6.5 phones posted on Oct 29th and as of end of Jan 2010 still having issues. Apart from that, nothing is really new in WinMo 6.5 - same old WinForm development with ugly UI centered around limited keyboard (as in key press even with no position information). So, despite that Windows Mobile Marketplace have just couple hundred apps and thus is ripe for plucking, despite that $99 might still entitle one to submit 5 apps for free (past Jan-01, 2010 expiration) as opposed to $99 per app submission, it ain't worth it developing for WinMo. (NB. students could submit apps for free and don't need to pay for Visual Studio, so what kind of applications we will have in the market place after a while?)

Bottom line is... Microsoft has very long way to go and it is not surprising that things are so pathetic. First they need to do a lot on SDK side and changes would need to be radical and it is doubtful they would be radical considering MS background and the way they implemented support for touch enabled devices. Second, considering the hassle developing for WinMo, application submission cannot cost $99 per submission. It won't work like that. Something like Blackberries $20 a pop in 10 quantities might work, $99 is plain way too expensive and until WinMo 7 gets proven track record would not work period.

Anyhow, porting games to WinMo would require porting CALayer and otherwise effort quite similar to Android port. Unlike Android thought there is no need for custom BoardView to handle touch, rather see hack above to handle touch events in the Form subclass (that would be an equivalent of Android activity). Re-read marketing for iPhone developers material and start with something trivial like FlashCards, Flickr (there is an app for posting to all blogs and alike, but no gallery app).

Sprint Blackberry cannot be activated on Boost CDMA without $50 plan or can they? May be by calling customer support once things settle down? Does it worth $90 when Touch Pro is 136? If BB cannot be activated on Boost CDMA consider Blackberry on TMO once 9 series would start selling in BestBuy.

Ti dolzhne blyad nam sdelat horosho, ne sebe, a nam

Boxee beta - first impressions.

Boxee beta still doesn't scrobble to last.fm. Nor does it work with MythTV. Nor can it do slideshow for local pictures. Nor do I see any activity published to my boxee account. Overall however boxee beta is a lot better then alpha that used to beat most media centers. Better, but still not good enough to become primary media center application. So would I buy boxee box?

GAE data storage

GAE data storage is pretty basic. Relationships are especially weak (guess this is inherited from both Java (JDO and such) and XML as a data store). Sure it works for simple object persistence but for more or less serious stuff it won't cut it. Let's consider simplest of them all bi-directional one-to-many. Data integrity aside it ain't worth it because of inflexibility. For managed one-to-many (read storing everything in one XML file) there are just two ways of resolving relationship - either doing nothing (default) or prefetching all (by setting @Persistent(defaultFetchGroup = "true")). That inflexibility comes at a cost of fetching all children just to add another one since there is no way of adding a child with some sort of foreign key and thus parent.getChildren().add(child) is the only way to add a child. Unmanaged one-to-many solve "foreign key issue", but come at a cost of updating parent any time child is added. The bottom line is - relationships should never be part of a class or data storage. They need to be modeled "orthogonally" to both objects and data storage in "data model" similar to EOF. When it comes to GAE and Java in general this implies same old - same old - very basic objects (XML documents as storage) with no JDO relationships. Relationships need to be modeled and implemented as part of data logic (read all relationship management needs to be implemented in getters and setters) - stone age or yet another way Java is Cobol of XXI century. N.B. Encoded string PK defined as

@PrimaryKey @Persistent(valueStrategy = IdGeneratorStrategy.IDENTITY) @Extension(vendorName="datanucleus", key="gae.encoded-pk", value="true") private String encodedKey; @Persistent @Extension(vendorName="datanucleus", key="gae.pk-name", value="true") private String pk;

are equivalent to Key and key generation using pk field as follows

@PrimaryKey @Persistent(valueStrategy = IdGeneratorStrategy.IDENTITY) Key id; ... this.id = KeyFactory.createKey(InAppPurchase.class.getSimpleName(), pk);

Hyundai is assaulting

Kia Rondo is the cheapest 7 seater that has the best value. It is a Hundai/Kia, i.e. similar to Toyota the car is no fun to drive, but overall package is actually quite good especially considering that after 40+% in initial depreciation in two years from $21K down to $11K it is the only econo-box that could be had with leather and V6. Rondo cloth interior has "autostain" feature, i.e. seats stains all by themselves, thus cloth seats are no go. Engine wise both I4 and V6 are adequate and more or less have the same fuel economy due to I4 coming with 4speed auto and V6 paired to 5speed auto. Base trims (I4+cloth) might not have 3rd row of seats, while V6+leather almost always come as 7 seater and this is the trim to get. Waranty wise, Kia 10yr/100K powertrain warranty is not transferable, yet 5yr/60K bumper to bumper is. 5/60K is standard these days, making any car build 2007 and later same as new built in 2006 (i.e. with only 3/36 that was the standard back then). Mazda 5 is the only other 7 seater that apart from minivan doors and useless 3rd row seats is pretty much the same, alas more expensive and available in leather only with later models... The rest of 7 seaters are minivan 200+ inches long (vs. 180- inches) or fullsize SUVs guzzling way bellow 20mpg. Both Rondo and Mazda 5 do 19/25/22mpg with Rondo's I4 outputing a hair more then Mazda 5 and doing a hair more mpg on a highway... Those fuel consumtion figures are your average for 5 seater compact wagons with germans giving more performace and fun for the same MPG and japanese giving 30+% more MPG for the same performance.

		L/W/H/ground	weight	cargo	turn	engine	MPG
2007 Rondo	$11K	179/71.7/65/6.1	3511	6.5/31.7	36.1	V6 192/184	18/26/21
2009 Rondo	$20K	179/71.7/65/6.1	3333	6.5/31.7	35.6	I4 162/164	19/26/22
2006 Mazda 5	$11.5K	181.5/69.1/64.2/5.5	4572	/44	34.8	I4 157/148	19/25/22
2000 323iT	$7.5K	176.3/68.5/55.5/	3351	25.7/	34.4	2.5L 170/181	18/26/21
2003 525iT	$11K	189/71/56.7/	3682	32.7/65	37.1	2.5L 184/175	17/24/19
2006 A4	$20K???	180.6/69.8/56.2/4.2	3671	27.8/59	36.4	2.0T 200/207	20/28/23
2009 Jetta TDI	$24K	179.4/70.1/59.2/5.4	4324	32.8/67	35.8	140/236	30/41/34
Insight	$20.5+K	172/66.7/56/	2723	15.9/32	36.1	98/123	40/43/41
2007 Fit	$12K	157.4/66.2/60/	2471	21.3???	35.6	109/105	28/34/31
Soul	$15K	161/70.3/63.4/6.5	2800	19.3/53	34.4	2L 142/137	24/30/26
2007 Vibe	$11K	171.9/69.9/62.2/5.9	2700	19.3/54	36.7	1.8 126/122	26/33/29
2009 Elantra Touring	$15K*	176/69.5/59.8/5.9	2937	24.3/65	34.2	I4 138/136	23/31/26

First let's look at the performance group split between BMW and Audi/VW (Volvo V50 has way to many problems to be considered). In summer 2006 VW finally fixed sludge in 1.8T by updating engine to 2.0T and making A4 worth the look. Alas 2006 A4 is still close to $20K then the rest are close to $10K and $20K could buy many new cars these days. On VW the same engine appeared only in 2009 Jetta together with new TDI, i.e. the priviledge to drive 2.0T would cost $20+K this way or another. In addition to 2.0T in 2009 VW started to put DSG on Jettas and Golfs making it the cheapest DSG out there, yet technology aside the value of DSG is questionable especially on Jetta with TDI where 5-speed should be optimal. This once again leaves BMW as the only performance wagons out there. 3-series is as good as it gets overall, and 5-series is the wagon when 3-series become too small. Suprisingly enough, 5-series is almost the same when it comes to performance/mpg as 3-series being way bigger car (well 19mpg combined is bellow 20mpg and 10% worse then 323iT with 21mpg). 2003 was the last year they sold 5-series wagon before the redesign and any 2000+ 5-series wagon is the best 5-series BMW, same as 4 cylinder E36 being the best 3-series. The bottom line however is that $10K and 20mpg used to move 5+luggage in style thanks to BMW, they still do, but also $10K+20mpg could now move upto 7 and have 2-3 years of warranty left. Also these days other options are not limited to VW with dirty cloth interior. Asian 5 seater wagons are not as fine as germans and hardly could be inspiring to drive, yet they offer far better fuel economy starting from 25-30+% more per gallon and all the way to twice or 100% more for Hybrid Insight. Insight is not quite the same class being smaller and crampier yet even with rear-seats up it could haul more then your average sedan (more then E36). Still the point about Fit being $5K cheaper and more capable for 10 mpg less is a good point. Those $5K even at $4/gallon would buy 1250 gallons of gas (or 3+ years of driving assuming 12K/year or <400 gallons/year), thus to catch up Insight should burn x/31-x/41=1250, or x=1250*41*31/(41-31)~160K miles... (the formula thus is miles_to_catch_up = $*mpg1*mpg2/$pg(mpg1-mpg2)...) Being twice as efficient and twice as expensive to catch up with 21 mpg E46, Insight should drive $10K*21*41/$4*20 or over 100K miles ($3 gallon makes it close to 150K miles, while $5 per gallon still makes it 86K). Clearly fuel-efficiency cannot be major factor in selecting asian econo-box. It gotta be capabilities per aquisition costs and this is what makes slightly used Hyundais and such far better buy then Hondas. Their initial depreciation makes their value shine even more then while new. Thanks to Big 3 crysis thus the only car that could approach Kias "value" when used are Pontiac Vibe. Yet they still have a long way to go when Elantra Touring goes for $15K new *after rebates ($1.5K cash and $1K marketing to dealer). Ultimately this race to the bottom should push 1-2 year old 5 door wagons value bellow $10K and at those price points Vibe vs. Elantra is worth a test drive despite being offered in single trim and cloth only. Overall, Soul appears to be a bastard child in this segment, neither cheap, nor offering utility, and thus test drives are

• Insight vs. Fit Sport manual - both are Honda, read too expensive.
• Rondo vs. Soul
• Vibe I1.8 vs. Elantra Touring. 1.8L Vibe might not be adequate, 2.4L GT could match up a hair more powerful Elantra.

Despite touching $10K with 2 yo cars, asian econo-wagons are no match to 6 years old BMW, so the only news is that there are 7 seaters now.

Mac Mini

Apple came out with the first Mini worth buying. Finally, base model has DVD writer. Alas, despite GF 9400 GPU, Apple again failed to sell me. Apart from even base Mini being over-priced $100 ($599 vs. $499 it should cost to be worthwhile), this time around I have no need for C2D based desktop... What I need is Ion based AppleTV for $199 and while it is not available probably I would roll my own.

In the simplest scenario 10.6x10.4x3.5200W PSU, PCI-e Low profile case for $38.99 (QC is reported to be bad) plus Zotac GF 9300 Mini ITX board for $145 with x16 PCI-e and everything else (but S-Video) including 11b/g is all I need ($184) to build mini PC 3" wider and deeper then Mini. Add to that 2.5GHz E5200 for $70 to match up new Mini at $39+$145+$70=$184+$70=$255 or better yet load it up with Q9550 (2.8GHz 2x6MB vs. 2.4GHz 2x2MB for Q6600) at $280 to make it fastest pre i5 Quad for $464 with CPU cost to be 60%. Newegg has Q9300 (2.5GHz 2x3MB) for $240, so give or take $40 with Q9550 being better option.

AMD CPU should be cheaper. Phenom 2 starts at $170 at MicroCenter and uses AM3 socket and Phenom 9600 is $100 2.3GHz at newegg after $20 rebate. Since AM2+ processors do not work on AM3 motherboards, depending on what type of GF 9400 mobo Zotac come up with for AMD (AM2+ vs. AM3), I could either use Athlon from HP and buy Phenom 9600 later or would have to buy both mobo and Phenom 2 at the same time. So AMD system would run me either $184 upfront, plus $100 later or $284 for 2.3GHz Phenom ($30 extra vs. base C2D) or $184+$170=$355 or $100 cheaper then Q9550.

Well, I have to wait for Zotac GF 9300 Mini ITX for AMD before building Mini PC... and may be by the time they come out with one NVidia would have ION out and old TV could be driven by ION based thingy for less then $200.

P.S. None of the Mini-ITX mobos have S-video out, and none but GF 9400 based have x16 PCI-e. Apple TV in addition to HDMI has just component video out, so only HP+X1550 combo could drive my old TV since it could only take S-video. My TV is showing its age.