They don't. The initials ESR identify Eric S. Raymond, while the acronym ESR usually refers to Effective Series Resistance, a property of capacitors. These are entirely different things!
The yearly average depends on the stage of the project. A young project with a still-growing code-base would likely show a huge average in the first few years, then the addition of more code after that would be much slower, with a lower average. And after 4 years, your 25k lines will be an average of 6250 per year.
Vista is definitely mature. Beside that, it is supposed to reach some kind of completion milestone soon, where one would hope there is no longer being thousands of lines of brand-new code still required to be added. Rather, it would be expected that the total to vary slightly as buggy code is being removed and replaced by less buggy code, with a slight slow increase.
Yes, I read the featured article afterwards, and they want something a lot more subtle than this, which is admittedly contrived. Anything containing something as obviously evil as gets() is probably disqualified from even entering the race.
I guess mixing fputs() and printf() like this makes the code look weird and that gives the game away.
In addition to the infamous gets() (not fgets(), which is a safe replacement for gets()) there is the call printf(stuf); where interesting things will happen if the string that was typed in contains percent signs directing printf() to try formatting or even writing to the stack in the vicinity of the call itself. This could be the return-address of the printf() call, or some other neighboring variables; and carefully crafted combination of excess bytes here could become interpreted as if it were executable code, and the machine could subsequently be pwnd. The post by whitenaga above gives an example.
My example was a bit contrived by the way, just a handful
of some known bad practices I banged together in five minutes...
From what I have seen from discussions on Groklaw, the subpoenas for HP and SUN is mostly a question of who all the people who have seen UNIX source at these companies are, and one possible answer might be "there are too many to count or list in a reasonable way". Which will be fine for IBM, since it allows them to confront SCO with the fact that UNIX was not ever a closely held trade secret. Also, it will raise the question of how come SCO is going after IBM and not HP or SUN or any of the other UNIX companies, considering that there must have been much of the same System V code behind HP-UX or Solaris as there is for AIX.
For MS and Baystar however, the subpoenas seem rather more ominous. IBM is looking for the control and the finances, and here is where things can become really "interesting" rather quickly. Who paid for the FUD and who decided it was an idea to be tried to begin with?
Anyway, IANAL, and I recommend going to Groklaw and looking there.
I read this thread, and then started enumerating the various computers around here... turned out to be more of them than expected. Now, only a few of these are actually on, and a couple others are so obsolete as to be nearly useless, but they all work - here goes, starting with the ones in front of me:
These are connected to the LAN and stay on most of the time. Yeah, there is the power bill, but the winters are cold here, and I can use the waste heat.
AMD Athlon running Linux (Red Hat 9), which I am writing this on.
AMD K6 running Linux (originally Slackware, but the kernel has been updated to 2.6.10) This also has Windows NT4 as another part of dual-boot but I haven't used that for ages. Mostly IRC, some browsing and development.
Pentium 2 running FreeBSD, this is for http and ssh and it also has a connection to a thermometer for the outside, whose readings can be obtained inside the LAN on port 9991.
Dell Poweredge running Fedora Core 4. I do 64-bit development work on this one through ssh or KVM extension, as the machine itself is too noisy to work with up close. I've hooked this up to the mains through its own energy meter, so I can see how expensive it actually is to run.
Another Pentium 2, a HP Vectra that I was given since it wouldn't run Windows 2000 anymore -- I wiped it, got Slackware 9 going on it, and I use it for microcontroller development and file storage. It runs nice and quiet, so it is a nice machine for desktop use.
A 486 sits in the basement, and records the inside temperature and the activity in the boiler room once per hour. This runs Linux Red Hat 6, and has done so for many years continually. Uptime is only been limited by the quality of the mains power...
Another Pentium 2 sits in the garage hooked up to a voltmeter and relays, and currently runs cron jobs for turning the outside lights on in the evenings and off in the morning, and every week it charges the battery of the old car that is parked there for the winter. The voltmeter allows me to check the battery voltage.
The following lot is turned on and off regularly, from once a day and down:
The Fujitsu Siemens portable. This is the main work machine, it runs Windows XP, only, and I use it daily for most work-related development work,
The portable Dell 610. Used to be the main machine until its hard disk failed. I replaced the hard disk, and I use it as a secondary work machine and sometimes data-logging in one of the cars. Dual-boot Windows 2000 and Linux, I mostly use Linux on it.
The HP9000-370. This is about 20 years old, runs HP-UX 7, and still sees some occasional use for data-logging or experiments. I have a 6-way relay controller using IEEE-488 for it, and it has 5 serial ports to which many interesting and useful units can be connected.
There are 4 more machines which were put together from bits and pieces left over from other, earlier, kit that had failed, which run Linux or FreeBSD, but I don't use them for anything at the moment.
Then there is the little skinny 386 that once was used for a BBS, now I got Minix to run on it, with a NE2000 network card, so even this old box makes it onto the LAN occasionally.
An old Pentium 1 portable computer with a broken screen hinge. Used to be the main work computer until about 2000, but has since been used for occasional looks at old code. Runs MS-DOS+Windows 3.11, Windows 95, or Windows NT 4 depending on which external hard disk is connected.
All the above machines are able to connect to the LAN. There are a few routers and switches and hubs here, all of the $20 SOHO variety, I don't bother enumerating these. All the network-enabled machines have names associated with them, and this gives them more of a presence. The others become anonymous like bulk equipment...
Only one of the remaining machines is used regularly.
The JDR-XT, my first computer actually, which I bought as a set of parts in 1986, and assembled. It has an EPROM burner which h
It doesn't seem that standardization on the language level is the problem. C, Perl, Python, C++, Java, C#, FORTRAN, Ada, and so on are just about as standardized as any screw thread such as UNC, UNF, Metric Fine, Metric Standard, Whitworth... In both sets there are plenty of options. The benefit of this standardization, that no-one is in much doubt of, is that companies or machine shops do not have to waste time on designing, testing and debugging their own language or screw threads, but can use whatever fits the job from the available selection, and put together software applications or machines or other structures.
And just like there are places where coarse or fine pitch threads are the best choice, so it is with programming languages. This analogy also suggests that one does not want to select too many different ones, but two or three seems to be better than one, and four or more seems excessive.
Please give an example of just one permanent law created due to what scientists have claimed.
How about Hooke's law on deformation of structural elements when external forces are applied to them, leading to legal requirements for civil engineers? Or the statistic observation on heating of electrical conductors from electrical currents, and on physiological effects of electricity on the human heart leading to the National Electrical Code?
Both physical laws are empirical and claimed as observations, and have limited applicability: Hookes Law about proportional deformation under load is only good until the loaded structure breaks down; likewise, the expectation of constant resistance or current-carrying capability breaks once the cable or connector has heated up enough to cause metallurgical changes in the material. Yet both are fundaments for important and long-lasting regulations.
Actually, the keyboard remapping is not on the same level as using different languages. They are more on the level of having to use the same old alphabet in a new sequence. Imagine a country or province where all signs, newspapers etc. had been printed with a specific transposition of letters, (say, rot13 as an example, but any of the other possible 26!-2 combinations would have been the same) and then having to quickly switch back and forth between them? You can probably do it with some practice, but in the meantime it has slowed you way down, for no reasonable purpose at all.
I find even the punctuation moving around between Norwegian and US standard keyboards to be bad enough when having to type the simplest of texts (text with only letters, digits, period, comma, and exclamation marks; the rest of the punctuation has moved around). It isn't like I cannot use it, but once I get to a point where there should be a question mark and i get an underscore instead, or a left parentheses and not the right one, I have to slow way down from the "typing while looking at the screen"-mode, where the thoughts, the fingers, the keyboard and associated circuitry run like a smooth-flowing pipeline, about as easy as talking, to "hunt-and-peck" mode, where the flow is more like the stop-and-go traffic of a city street grid.
This has nothing to do with language, but all to do with the path from brain to machine.
There weren't much in the way of "personal" computers used in business. Before the microprocessor-based units, the "business computers" were larger units that had terminals connected to them, and which were what the users used. As far back as the late 1970s, I remember seeing ads for various minicomputers and office computers, but these were way too expensive for any personal use, even the smaller ones using 8 inch disks and CP/M. A few years later, I got to work with one of these, a 16-bit device, roughly the size and shape of a small fridge, called "Naked".
Later, in the 1980s, we had all the "workstations" with different varieties of unix, and these were personal to some extent, in that a designer or engineer could use it as his or her own. The IBM PC and its MS-DOS and CGA monitor with 4 colors and blocky resolution looked like a toy in comparison to the HP workstations with their 19 inch color monitors running X-window or the IBM RTs running AIX and also having nice color screens (14 or 15 inch iirc) and plenty of colors. Of course, the price difference reflected their playing in different leagues.
Once the PC/AT and the EGA graphics came out, the PCs became sufficiently useful so as to replace the nice, capable, heavy, and expensive workstations, many of which were 32-bit systems with processors such as the 68020 and its successors.
Then the successors to the IBM-PCs, the 32-bit 386 and 486 machines came, but it was not until some time into the 1990s that the most widely used operating system, the ones from Microsoft, actually used all these 32 bits. The reason why MS-DOS stayed around for so long was that all the PCs successors were compatible with the previous ones, and MS-DOS, and in particular, with Windows 3 running on top of it, remained sufficiently useful. Linux wasn't that well known before about 1994 either, even though it started out as a 32-bit operating system, and the BSDs were hung up in a dispute with the old AT&T for much of the 1990s. Most people then, as now, don't care about the operating system as long as it works.
This appears to be a recipe for what a former colleague called "data death", basically being drowned in more data than it would ever be possible to figure out how to make use of.
Even if they only log e-mails and TCP connections, what will they see? That over some period of time I get a few thousand spam mails from miscellaneous sources and a few hundred mails from various colleagues, family members, IEEE, and vendors? Or that I ssh into a machine here from work, and I have thousands of idiots banging away with their dictionary attack attempts on the ssh server here from all over the world? Or that a similar number of idiots are trying to crack the webserver by looking for files there that don't even exist? Or that I have long-lasting connections to two different IRC networks? Then occasional MSN connections; although I don't use that much, I know people with teenage kids who are on this all the time. Still, there will be no indication about what the contents of all this are. How the spam divides into money scams, pill-pushers, penis-improvements, breast-changes (yes, the spammers bother with sending both of these) and sales of fake Rolexes?
And then, the WEB, http, Port TCP 80, that part of the internet commonly confused with the internet at large. I go to read one of the general newspapers, and you can bet that there will be one indication of me looking at, say, http://www.vg.no/ and then a dozen looks at all the ad providers... Yeah, so I read VG on the network, like a couple other million others in this country. That's gonna be a helluva lot of mostly totally interesting data. Of course, not everything is quite as irrelevant as that: it can also reveal which bank I am using, and how often. Does anyone ever have a good reason to get to know this outside of a search warrant? I think not.
And what about UDP, such as is used for VOIP? that doesn't really make "connections" though it is possible, by inspecting the packages, to see the source and target address. But all it will say a million times over is that I have sent and received UDP packages from the VOIP provider's server... during some particular time period. Or if I used something like Skype, it would be to whoever else I talked to, like voice phone or SMS. Ere this legislation, there were requirements for a search warrant or surveillance warrant before this information could be made available.
So they will be able to see "when" and for the http at least, "with whom", given some filtering. Imagine that the post-office kept a journal of who posted and received mail, and when this happened? Same thing, just more work-intensive and expensive, and with the same slew of uninteresting info being produced. Though I think there are constitutional laws in most of the countries prohibiting this kind of logging, if it was ever thought of as not being just too unreasonable in terms of cost or manpower...
Thing is, the filtering can only make sense once the police or whoever is supposed to be able to obtain this information, knows what to look for. In the meantime there are going to be many terabytes per person per year -- there are 450 million people in the EU, even if only two-thirds were active, we're talking the need for storing on the order of 10^21 bytes worth of logs for the lot, for the 2-year storage time. I don't know what the total output of mass storage production is per year, but this would be some 10^12 DVDs, or 10^13 CDs, or 10^10 harddrives of 100 GB each... I don't know if such a large number of hard drives ever has been produced to date.
And never mind who is going to have to pay for this... Whether it is taxpayers though government or ISP's customers, it is going to be an enormous economic brake on all kinds of business when all money goes to purchasing and maintaining storage units and searching through them. And to what purpose? to find out who is the terrist before they actually blow something up? Not much use if you end up destitute in the process, only with a filled storage
So basically, from the featured article, an "amendment" to an economic bill can result in rather severe changes to how the various parts of the MA government can operate? Somehow this seems familiar, we've seen these things happen before in the EU and elsewhere in the US when certain big mono- and oligopolists wants something which is not quite within scope of the currently proposed legislation, something that would be more obviously suspect if proposed as a separate bill.
Perhaps it is time for general ban "riders" and "amendments" that change the overall meaning of some proposed regulation, unless their consequences have been evaluated, or perhaps just an overall ban: if a bill is to be changed or augmented, it has to be cancelled and reintroduced from scratch. Something needs to be done about rampant law-making in general, this part of governments have been allowed to run riot and overwhelm the judicial processes. In MA as elsewhere.
Of course, this would just be another law, adding to the mess...
In theory, anything with a microcontroller could be reprogrammed. In practice there tends to be a number of obstacles put in the way of performing re-programming, in these cases where re-programming never is expected to be needed, or when re-programming is not desired for whatever reason. Mouse and garage-door opener controllers are typical examples of mask-programmed controllers, where there is a large volume and a simple function that never needs to be changed in the field; if there are changes these are applied at the factory, and a new mask is laid for the new revision or model of controller.
Other systems may use one-time-programmable chips, and most of these have various kinds of "security bits" that effectively slams the door on the possibility of reading the existing program and changing it. Typically something done in order to retain trade secrets.
Now, there are also field-programmable units, whether memories or erasable and re-writeable controller, with some kind of EEPROM memory in them. Even if they can be erased and re-programmed, these still tend to have some kind of security mechanism for the benefit of those wanting to keep their trade secrets.
I do expect that Microsoft has not made it easy to reprogram or even inspect the contents of these memory areas that hold the key for the unit. And it's not like there'd be a separate 24C01 memory chip with an I2C-bus interface holding the secret key, we can expect that there are some nonvolatile bytes of memory safely tucked away inside the chip.
Very likely, this memory is designed as externally write-only-once, so that once the key is written it can never be either read or rewritten. This resembles the region coding change limitation on DVD-drives, where the region code may be changed N times and it eventually sticks at the last one. Reduce N to 1 here, don't implement any way of externally reading the value, and there it is.
That does not preclude the possibility of overall testing of the external response to stimuli and deducing the internal secrets; but it does make this job quite a bit harder. After all, the device must eventually be able to run code from an external data source, as opposed to embedded controllers that have a fixed program that hardly ever changes.
As for a wall-socket style, I guess the same kind as are commonly seen in automobiles, the cylindical, "cigar-lighter"-style, could be considered as a standard.
However, the problems still remain with the various connectors that go into the equipment at the other end of the cable going from the wall-wart, and all the various voltages and polarities, as well as the maddeningly large variety of similar-looking but not quite compatible plugs and sockets (2.1 mm, vs 3/32", vs 2.5 mm center pin diameters for example)
There's also the nontrivial problems of implementing large scale low-voltage power distribution grids -- in cars, where the most common implementations of such low-voltage grids are found, the distances are a lot shorter than in any house, and this makes for a lot less of a risk of undesired voltage drops and interference from other unknown devices. There are motors and other sources of generated noise in the car's electrical grid, but these are to a large degree well-known, and the remedies are as well.
In the house in contrast, the distances are a lot larger, leading to all sorts of possibilities of induced noise voltages from nearby equipment, such as the regular AC mains, The problem comes from the combination of low voltage and large current, and that a noise spike from a current being interrupted, as when a heater or motor is turned on or off, will be relatively bigger for a low-voltage system. At 120 or 240 V, a spike of 4V up or down is of insignificant size: the same 4 V spike or droop in a 12 V system will become really noticeable, whether it is your lights dimming or your TV-set-top-box rebooting...
There is also the matter of requiring a lot heavier gauge wires if the low voltage outlets are to supply any large amount of power, though at the power levels that we are seeing with wall-warts, this is on the order of tens of watts. A 16 A circuit can deliver nearly 200 W at 12 V, and that could possibly replace a lot of wall-warts... thing is still, if they are replaced by automotive-standard ones, where will the net savings be? Because all equipment has their own voltage and current ratings, I've got various wall-wart driven things here that range from 3 V DC to 30 V DC, and then there are 9 V AC, 24 V AC, and center-tapped 2x8 V AC (this last one is a brick on the cable)... and no DC power distribution networks is going to be any improvement for these latter ones.
I once had a job at a place where they had common power supplies and low-voltage distribution to panels in the labs. The offerings included 5 V DC, 12 V DC, 120 V 3-phase 400 Hz AC, and a few others, but noise from equipment in one lab would appear in the power supplied to any of the others, so those low-voltage DC supplies turned out to be not really useful at all.
Some others have given some really good explanations on most of these questions, I'll try explaining a little more about the way the system deals with hardware.
The operating system kernel is this part of it that runs all the time and takes care of the scheduling of the various processes and of servicing the requests that these have for access to disks, keyboard, mouse, screen, serial ports, network ports, printers and so on. Since there is a lot of different varieties of these, the operating system uses a set of suitable drivers for each device. The devices and their associated drivers are of three major kinds: there are character devices, that work with streams of characters, such as the keyboard or the screen or the printer, and there are block devices such as disks, that deal with large chunks of data. The third kind of device are network interfaces that are not classified as traditional block or character devices at all.
The block and character devices are identified in the systems with their "major number". In older unix systems, there was two master tables of these device numbers, that indicated that for example character device number 1 would be a console, block device number 7 would be some kind of disk and so on.
On Linux systems, the same idea of major numbers are used, but the device driver programs usually request their own major numbers themselves, and the kernel builds these tables on boot, and we can see what this becomes through the pseudo-file/proc/devices.
Programs in the "user-space", that is all programs that we normally run as processes ourselves, have to access these devices through "node files", normally located in the/dev directory. When reading or writing to a device through these files, the operating system routes the requests for reading and writing to the appropriate driver since it knows the major number. Each driver provides a read() and a write() function to the kernel code that it calls for the device-driver to do its job. Both character and block devices may be read and written as if they were disk files, the block devices hold filesystems, and can additionally be mounted, and thus made accessible through a directory somewhere in the system.
The node files in/dev indicate a filename of a device, its major number, its minor number (this is used to tell several different devices of the same type apart), and of course whether this is a block or character device. On a long listing, (ls -l) the prefix shows as b or c for block or character device. Like other files, these may have read/write permissions set to control who may read or write the files. Most of them are owned by root. It is possible to create such node-files elsewhere in the file system, but that tends to be frowned upon.
brw-rw---- 1 root disk 3, 0 Jan 30 2003 hda brw-rw---- 1 root disk 3, 1 Jan 30 2003 hda1 brw-rw---- 1 root disk 3, 2 Jan 30 2003 hda2 crw-rw-rw- 1 root uucp 4, 64 Feb 21 2005 ttyS0 crw-rw-rw- 1 root uucp 4, 65 Feb 21 2005 ttyS1
In the above, very short, excerpt from/dev on this machine, we see the hard disk is block device major number 3, and the serial ports are character devices major number 4.
There is usually a whole slew of such node files under/dev, and they come in long series of names, beginning with some short prefix such as tty, pty, sda, hda, fd and so on. Generally, these prefixes follow a convention:
tty are serial lines, terminals, or consoles, (the abbreviation comes from "teletype"); pty are "pseudo-tty" devices, basically a kind of pipe that look much the same as a real tty to the programs reading and writing to them. Another character device of interest is the lp series, these are the parallell ports as used for printers.
There would still be a need for a DC-DC conversion, since the components inside the servers will need various DC voltages, such as 3.3V, 5V, 12V, -5V and -12V, that are different from the battery voltage level, whether this is 12V, 24V, 48V, 120V, or whatever. Telco kit running on 48V DC contains similar DC-DC converters.
Such a DC-DC conversion already takes place inside the usual computer power supply. The first thing that happens to the mains power going in is that is being rectified, before a high-frequency oscillator circuit generates an AC voltage again, which then is rectified and produces the various voltages required by the server components.
As noted, low voltage DC implies fat and expensive conductors. So we could feed our servers with a higher DC voltage, and get away with reasonably sized conductors?
High voltage DC, while reducing the required current for the same power, still is rather nasty with its ability to sustain hot arcs. This makes for very expensive switches and fuses that won't just melt upon first attempt at disconnecting. I suspect one could use some variant of solid-state switches for breaking the load, combined with low rating switches to isolate the load electrically once the current is low, but this is the kind of complicated arrangement you'd expect to find in a power substation where there are several kilovolts of AC. Or maybe a rectifier station for a railway or streetcar system that runs on DC. But not the infrastructure of a server room...
I guess we could have individual UPS-supply combinations for each server machine, much like we have with batteries in notebook computers, but the fire-codes may require that all power can be switched off to the machines in the server room.
Then there's the economies of scale. Mains-powered equipment is readily available and have reasonably efficient power supplies. Distributing DC at the same power level (say a few kW) will require much more expensive components, whether these are sturdier switches or fatter conductors. Switching supplies are widely used and they are quite efficient, whether they have to convert AC to DC, DC to DC, or DC to AC, makes little difference; the energy savings and money savings will be in the cables and switches.
Pay for content. The revolution with the Web is that there is no limitations or anyone controlling the contents there. It used to be, with television, radio, and books, that only the select few producers were able to reach a large audience. Now this has changed to be determined by what you, YOU the reader and potential producer, have to say, and whether, or rather to what extent anyone's interested in it. Now anyone can read, and thanks to Google, anyone can find something they're looking for (as in it may not be what they want, but it will be what they need).
Had the web been created today by any media corporation or association of these, it would have been just another variation on the pay-for-content and "We produce, you consume" theme that is the bread and butter of the media companies today. They do not want to have any competition. And they do not want to surrender their control of the distribution channels.
They, Microsoft, also get to restrict what OS/filesystem gets installed on the system too. So while the number of GNU/Linux users is still ONLY in the 10-20 Million, their voices need to be pretty loud on this one. IMO.
That MS may restrict what can be run under their OS might be problematic enough, though possibly within their rights. Just look at their X-box. Of course, what will happen is that the developers will start using and advocating other platforms to an even larger degree.
And we can expect that MS forcing the production of Windows-only non-MS "PC" hardware (the Xbox is an MS loss-leader, and their baby) will go over like a lead balloon. Though I see the movements in that direction. Closed hardware is already endemic, all sorts of non-open specs apply to wireless network cards and screen cards. The extension to closing the rest of the machine is a reasonable extrapolation.
However, here is where someone will cry foul... what with MS being already a monopoly and using their operating system to entrench this? Yeah right, that is going to go really well with the courts, in either the US or Europe or the Far East.
And then consider also Massachusetts who just decided to go with all open-standard documents and open standard only -- this could already turn into a quite complicated issue for Microsoft, and if MS decide to make things difficult for the alternate operating systems, anything might happen, from them becoming just irrelevant for business use, to the re-opening of the monopoly case.
Add a 5th, and very big reason: People will continue to buy Windows because it comes with their new machine.
Even as easy as pre-installed Windows is to get running on a brand-new machine, I still get occasional phone calls from people whose newly purchased machine gives them trouble accessing the Internet, since the supplied modem cable was plugged into one of the sockets on the switch in the LAN...
Furthermore, the awareness of alternate operating systems or applications varies, many of them only know what Linux is based on what I have been telling them. These are not IT professionals, they see the computer as some kind of information-appliance, somewhat similar in function and purpose to some other well-known appliances such as typewriters, TV, filing cabinet and to some extent CD players. Like these, it can break (and they tend to be afraid of doing anything that could make that happen), and need to be fixed or replaced.
It is agurktid or "cucumber season" here in Norway. Whether this is from the tendency for the news bulletins to have about 3% substance, or it is from the period when all the papers had to write about was the size of the cucumbers on the town market... I don't know, but certainly, the first interpretation seemed to fit better in this case.
And in a similar vein, Qantas had renamed their purser to "Customer service representative". Now, to me that sounds like someone sitting behind a desk at the terminal rather than someone onboard a plane.
But some marketer probably came up with the change; even though these customers are a captive audience by the time they meet this "representative".
"Gadget porn" indeed. Gives the sleazy-ish connotations and all, and the readings on me troll-o-meter go to the middle of the scale. Gets me right into debunking mood. There is a number of correct statements, but the overall impression is pseudo-science, and subjective impressions of "better sound".
The starting premise, that manufacturers use the least expensive components they can get away with is no big news. This has been considered good engineering just about forever: use those parts that are good enough technically and least costly. The key is "good enough", though for what? The subsequent discussion on audio and video quality just indicates that the player is working as it should. As to distortion, there is only the subjective opinion that the cymbals don't sound as clear as they should, but there is no further indication of whether this comes from the player or the amplifier chain following.
Then the discussion proceeds to take apart and redo the power supply... Not components in the signal path, but the power supply. A switching power supply that is powered direct from the mains, with X and/or Y-rated capacitors and inductors in the power entrance, and a somewhat carefully orchestrated feedback loop that causes its components to oscillate and generate the correct DC voltages for the electronics. Apparently, (and fortunately) most of the really important components are surface-mount and thus not amenable to this kind of tinkering. The argument goes on that this more "stiff" power is needed for more accurate bass-response. Hello? this isn't a 30 Watt tube-amplifier we are talking about here, where such an argument might hold, but something that puts out a few tens of milliwatts of power into maybe 600 Ohms at the most. Unless the power supply is pathologically feeble, this is really just -- marketing-speak, to use a polite term.
However, even if putting in larger filter electrolytic capacitors might be harmless, the replacement of the X-rated capacitor with an "Auricap" which is evidently NOT X-rated, sounds dodgy, as in potential fire hazard. The Auricaps seem to be marketed as non-electrolytic replacements for electrolytic capacitors in the signal path, and might do a fine job here, but we're not expecting any of our precious signal to enter the mains are we?
The fact that there is an X-rated capacitor there at all and not just a cheaper one, is that this is sitting across the mains voltage, and has to conform to specific requirements from the UL, CSA, TUV and so on, lest they refuse to List or certify the equipment for sale. And probably more important, that the component fails safe and does not start a fire.
Googling for "aurocaps" shows several sites catering to the same merry lot who depends on "oxygen-free" cables, and praise the virtues of the gold-plated RCA connectors... There is a reason why professional kit uses XLR or BNC connectors, or even 1/4inch jacks by the way, and it has to do with mechanical and electrical stability and shielding, not any magic properties of gold or nickel that makes one good and the other bad.
It goes on about replacing more of the bypass capacitors in the digital processing section, and mentions the possibility of clock jitter. Technically correct. But no quantitative information, no measurements done on a distortion analyzer or even a picture of the signal on an oscilloscope. Just all this non-scientific hand-waving that if we put in more expensive capacitors the sound will be better.
Finally, op-amps and possible replacements. Again, the observation is that the amplifiers are low cost, and obviously we could put in better and more expensive ones. Low cost is not the same as crummy; had the manufacturer put in really bad ones, everyone would have heard. Again, it is a matter of good enough, though the only parameter that might make a difference would be the noise of the amplifier. Unless they are really atrocious (with obvious effects on sales), gain and slew-rate would not matter, except for marketing purposes.
My guess for what might constitute the perceived "improvement" in this case, is that the frequency response of the audio chain has changed, boosting the higher frequencies, and thus made the modified unit appear to sound better.
If one were to make a replica of a 57 bel air, it would probably be a good idea to use an existing, more recent set of parts for the engine, transmission, steering, and braking systems. There are all sorts of safety and efficiency considerations here, and quite a number of improvements have taken place in the nearly 50 years since the original (disc brakes, dual brake circuits, fuel injection, rack-and-pinion steering, all come to mind as useful improvements over the original).
Most of the body parts like quarter panels and door bodies, floor, hood, etc. would probably best be made out of sheet metal, suitably bent or pressed. Other structural parts probably could be best made out of pipe stock. There still exist drawings of these components, and it wouldn't be impossible for some metal shop to make them, even considering adjustments for the different mechanical parts.
But smaller, irregularly-shaped parts, such as door handles, instrument panel, light-fixtures and much of the cast chromed parts, could very well be made with some of these new methods.
The resulting vehicle would perhaps become a '07 Bel Air?:)
Another thing is that these fabrication methods can be used for making new specialty engine or other parts for old cars in general, where no modern useable equivalent is available.
Slashdot Mirror
They don't. The initials ESR identify Eric S. Raymond, while the acronym ESR usually refers to Effective Series Resistance, a property of capacitors. These are entirely different things!
The yearly average depends on the stage of the project. A young project with a still-growing code-base would likely show a huge average in the first few years, then the addition of more code after that would be much slower, with a lower average. And after 4 years, your 25k lines will be an average of 6250 per year.
Vista is definitely mature. Beside that, it is supposed to reach some kind of completion milestone soon, where one would hope there is no longer being thousands of lines of brand-new code still required to be added. Rather, it would be expected that the total to vary slightly as buggy code is being removed and replaced by less buggy code, with a slight slow increase.
Yes, I read the featured article afterwards, and they want something a lot more subtle than this, which is admittedly contrived. Anything containing something as obviously evil as gets() is probably disqualified from even entering the race.
I guess mixing fputs() and printf() like this makes the code look weird and that gives the game away.
In addition to the infamous gets() (not fgets(), which is a safe replacement for gets()) there is the call printf(stuf); where interesting things will happen if the string that was typed in contains percent signs directing printf() to try formatting or even writing to the stack in the vicinity of the call itself. This could be the return-address of the printf() call, or some other neighboring variables; and carefully crafted combination of excess bytes here could become interpreted as if it were executable code, and the machine could subsequently be pwnd. The post by whitenaga above gives an example.
My example was a bit contrived by the way, just a handful of some known bad practices I banged together in five minutes...
#include <stdio.h>
main()
{
char stuf[80];
while(1)
{
fputs("Enter something: ", stdout);
fflush(stdout);
gets(stuf);
fputs("You have entered ", stdout);
printf(stuf);
}
}
silly (and looks innocent enough) but closer inspection will reveal nastyness...
For MS and Baystar however, the subpoenas seem rather more ominous. IBM is looking for the control and the finances, and here is where things can become really "interesting" rather quickly. Who paid for the FUD and who decided it was an idea to be tried to begin with?
Anyway, IANAL, and I recommend going to Groklaw and looking there.
I read this thread, and then started enumerating the various computers around here ... turned out to be more of them than expected. Now, only a few of these are actually on, and a couple others are so obsolete as to be nearly useless, but they all work - here goes, starting with the ones in front of me:
These are connected to the LAN and stay on most of the time. Yeah, there is the power bill, but the winters are cold here, and I can use the waste heat.
AMD Athlon running Linux (Red Hat 9), which I am writing this on.
AMD K6 running Linux (originally Slackware, but the kernel has been updated to 2.6.10) This also has Windows NT4 as another part of dual-boot but I haven't used that for ages. Mostly IRC, some browsing and development.
Pentium 2 running FreeBSD, this is for http and ssh and it also has a connection to a thermometer for the outside, whose readings can be obtained inside the LAN on port 9991.
Dell Poweredge running Fedora Core 4. I do 64-bit development work on this one through ssh or KVM extension, as the machine itself is too noisy to work with up close. I've hooked this up to the mains through its own energy meter, so I can see how expensive it actually is to run.
Another Pentium 2, a HP Vectra that I was given since it wouldn't run Windows 2000 anymore -- I wiped it, got Slackware 9 going on it, and I use it for microcontroller development and file storage. It runs nice and quiet, so it is a nice machine for desktop use.
A 486 sits in the basement, and records the inside temperature and the activity in the boiler room once per hour. This runs Linux Red Hat 6, and has done so for many years continually. Uptime is only been limited by the quality of the mains power...
Another Pentium 2 sits in the garage hooked up to a voltmeter and relays, and currently runs cron jobs for turning the outside lights on in the evenings and off in the morning, and every week it charges the battery of the old car that is parked there for the winter. The voltmeter allows me to check the battery voltage.
The following lot is turned on and off regularly, from once a day and down:
The Fujitsu Siemens portable. This is the main work machine, it runs Windows XP, only, and I use it daily for most work-related development work,
The portable Dell 610. Used to be the main machine until its hard disk failed. I replaced the hard disk, and I use it as a secondary work machine and sometimes data-logging in one of the cars. Dual-boot Windows 2000 and Linux, I mostly use Linux on it.
The HP9000-370. This is about 20 years old, runs HP-UX 7, and still sees some occasional use for data-logging or experiments. I have a 6-way relay controller using IEEE-488 for it, and it has 5 serial ports to which many interesting and useful units can be connected.
There are 4 more machines which were put together from bits and pieces left over from other, earlier, kit that had failed, which run Linux or FreeBSD, but I don't use them for anything at the moment.
Then there is the little skinny 386 that once was used for a BBS, now I got Minix to run on it, with a NE2000 network card, so even this old box makes it onto the LAN occasionally.
An old Pentium 1 portable computer with a broken screen hinge. Used to be the main work computer until about 2000, but has since been used for occasional looks at old code. Runs MS-DOS+Windows 3.11, Windows 95, or Windows NT 4 depending on which external hard disk is connected.
All the above machines are able to connect to the LAN. There are a few routers and switches and hubs here, all of the $20 SOHO variety, I don't bother enumerating these. All the network-enabled machines have names associated with them, and this gives them more of a presence. The others become anonymous like bulk equipment...
Only one of the remaining machines is used regularly.
The JDR-XT, my first computer actually, which I bought as a set of parts in 1986, and assembled. It has an EPROM burner which h
And just like there are places where coarse or fine pitch threads are the best choice, so it is with programming languages. This analogy also suggests that one does not want to select too many different ones, but two or three seems to be better than one, and four or more seems excessive.
The election is on September 17. This means that the results will be in just in time for September 19, The Pirate Day. :)
How about Hooke's law on deformation of structural elements when external forces are applied to them, leading to legal requirements for civil engineers? Or the statistic observation on heating of electrical conductors from electrical currents, and on physiological effects of electricity on the human heart leading to the National Electrical Code?
Both physical laws are empirical and claimed as observations, and have limited applicability: Hookes Law about proportional deformation under load is only good until the loaded structure breaks down; likewise, the expectation of constant resistance or current-carrying capability breaks once the cable or connector has heated up enough to cause metallurgical changes in the material. Yet both are fundaments for important and long-lasting regulations.
I find even the punctuation moving around between Norwegian and US standard keyboards to be bad enough when having to type the simplest of texts (text with only letters, digits, period, comma, and exclamation marks; the rest of the punctuation has moved around). It isn't like I cannot use it, but once I get to a point where there should be a question mark and i get an underscore instead, or a left parentheses and not the right one, I have to slow way down from the "typing while looking at the screen"-mode, where the thoughts, the fingers, the keyboard and associated circuitry run like a smooth-flowing pipeline, about as easy as talking, to "hunt-and-peck" mode, where the flow is more like the stop-and-go traffic of a city street grid.
This has nothing to do with language, but all to do with the path from brain to machine.
Later, in the 1980s, we had all the "workstations" with different varieties of unix, and these were personal to some extent, in that a designer or engineer could use it as his or her own. The IBM PC and its MS-DOS and CGA monitor with 4 colors and blocky resolution looked like a toy in comparison to the HP workstations with their 19 inch color monitors running X-window or the IBM RTs running AIX and also having nice color screens (14 or 15 inch iirc) and plenty of colors. Of course, the price difference reflected their playing in different leagues.
Once the PC/AT and the EGA graphics came out, the PCs became sufficiently useful so as to replace the nice, capable, heavy, and expensive workstations, many of which were 32-bit systems with processors such as the 68020 and its successors.
Then the successors to the IBM-PCs, the 32-bit 386 and 486 machines came, but it was not until some time into the 1990s that the most widely used operating system, the ones from Microsoft, actually used all these 32 bits. The reason why MS-DOS stayed around for so long was that all the PCs successors were compatible with the previous ones, and MS-DOS, and in particular, with Windows 3 running on top of it, remained sufficiently useful. Linux wasn't that well known before about 1994 either, even though it started out as a 32-bit operating system, and the BSDs were hung up in a dispute with the old AT&T for much of the 1990s. Most people then, as now, don't care about the operating system as long as it works.
Even if they only log e-mails and TCP connections, what will they see? That over some period of time I get a few thousand spam mails from miscellaneous sources and a few hundred mails from various colleagues, family members, IEEE, and vendors? Or that I ssh into a machine here from work, and I have thousands of idiots banging away with their dictionary attack attempts on the ssh server here from all over the world? Or that a similar number of idiots are trying to crack the webserver by looking for files there that don't even exist? Or that I have long-lasting connections to two different IRC networks? Then occasional MSN connections; although I don't use that much, I know people with teenage kids who are on this all the time. Still, there will be no indication about what the contents of all this are. How the spam divides into money scams, pill-pushers, penis-improvements, breast-changes (yes, the spammers bother with sending both of these) and sales of fake Rolexes?
And then, the WEB, http, Port TCP 80, that part of the internet commonly confused with the internet at large. I go to read one of the general newspapers, and you can bet that there will be one indication of me looking at, say, http://www.vg.no/ and then a dozen looks at all the ad providers... Yeah, so I read VG on the network, like a couple other million others in this country. That's gonna be a helluva lot of mostly totally interesting data. Of course, not everything is quite as irrelevant as that: it can also reveal which bank I am using, and how often. Does anyone ever have a good reason to get to know this outside of a search warrant? I think not.
And what about UDP, such as is used for VOIP? that doesn't really make "connections" though it is possible, by inspecting the packages, to see the source and target address. But all it will say a million times over is that I have sent and received UDP packages from the VOIP provider's server... during some particular time period. Or if I used something like Skype, it would be to whoever else I talked to, like voice phone or SMS. Ere this legislation, there were requirements for a search warrant or surveillance warrant before this information could be made available.
So they will be able to see "when" and for the http at least, "with whom", given some filtering. Imagine that the post-office kept a journal of who posted and received mail, and when this happened? Same thing, just more work-intensive and expensive, and with the same slew of uninteresting info being produced. Though I think there are constitutional laws in most of the countries prohibiting this kind of logging, if it was ever thought of as not being just too unreasonable in terms of cost or manpower...
Thing is, the filtering can only make sense once the police or whoever is supposed to be able to obtain this information, knows what to look for. In the meantime there are going to be many terabytes per person per year -- there are 450 million people in the EU, even if only two-thirds were active, we're talking the need for storing on the order of 10^21 bytes worth of logs for the lot, for the 2-year storage time. I don't know what the total output of mass storage production is per year, but this would be some 10^12 DVDs, or 10^13 CDs, or 10^10 harddrives of 100 GB each... I don't know if such a large number of hard drives ever has been produced to date.
And never mind who is going to have to pay for this... Whether it is taxpayers though government or ISP's customers, it is going to be an enormous economic brake on all kinds of business when all money goes to purchasing and maintaining storage units and searching through them. And to what purpose? to find out who is the terrist before they actually blow something up? Not much use if you end up destitute in the process, only with a filled storage
So basically, from the featured article, an "amendment" to an economic bill can result in rather severe changes to how the various parts of the MA government can operate? Somehow this seems familiar, we've seen these things happen before in the EU and elsewhere in the US when certain big mono- and oligopolists wants something which is not quite within scope of the currently proposed legislation, something that would be more obviously suspect if proposed as a separate bill.
Perhaps it is time for general ban "riders" and "amendments" that change the overall meaning of some proposed regulation, unless their consequences have been evaluated, or perhaps just an overall ban: if a bill is to be changed or augmented, it has to be cancelled and reintroduced from scratch. Something needs to be done about rampant law-making in general, this part of governments have been allowed to run riot and overwhelm the judicial processes. In MA as elsewhere.
Of course, this would just be another law, adding to the mess...
Other systems may use one-time-programmable chips, and most of these have various kinds of "security bits" that effectively slams the door on the possibility of reading the existing program and changing it. Typically something done in order to retain trade secrets.
Now, there are also field-programmable units, whether memories or erasable and re-writeable controller, with some kind of EEPROM memory in them. Even if they can be erased and re-programmed, these still tend to have some kind of security mechanism for the benefit of those wanting to keep their trade secrets.
I do expect that Microsoft has not made it easy to reprogram or even inspect the contents of these memory areas that hold the key for the unit. And it's not like there'd be a separate 24C01 memory chip with an I2C-bus interface holding the secret key, we can expect that there are some nonvolatile bytes of memory safely tucked away inside the chip.
Very likely, this memory is designed as externally write-only-once, so that once the key is written it can never be either read or rewritten. This resembles the region coding change limitation on DVD-drives, where the region code may be changed N times and it eventually sticks at the last one. Reduce N to 1 here, don't implement any way of externally reading the value, and there it is.
That does not preclude the possibility of overall testing of the external response to stimuli and deducing the internal secrets; but it does make this job quite a bit harder. After all, the device must eventually be able to run code from an external data source, as opposed to embedded controllers that have a fixed program that hardly ever changes.
However, the problems still remain with the various connectors that go into the equipment at the other end of the cable going from the wall-wart, and all the various voltages and polarities, as well as the maddeningly large variety of similar-looking but not quite compatible plugs and sockets (2.1 mm, vs 3/32", vs 2.5 mm center pin diameters for example)
There's also the nontrivial problems of implementing large scale low-voltage power distribution grids -- in cars, where the most common implementations of such low-voltage grids are found, the distances are a lot shorter than in any house, and this makes for a lot less of a risk of undesired voltage drops and interference from other unknown devices. There are motors and other sources of generated noise in the car's electrical grid, but these are to a large degree well-known, and the remedies are as well.
In the house in contrast, the distances are a lot larger, leading to all sorts of possibilities of induced noise voltages from nearby equipment, such as the regular AC mains, The problem comes from the combination of low voltage and large current, and that a noise spike from a current being interrupted, as when a heater or motor is turned on or off, will be relatively bigger for a low-voltage system. At 120 or 240 V, a spike of 4V up or down is of insignificant size: the same 4 V spike or droop in a 12 V system will become really noticeable, whether it is your lights dimming or your TV-set-top-box rebooting...
There is also the matter of requiring a lot heavier gauge wires if the low voltage outlets are to supply any large amount of power, though at the power levels that we are seeing with wall-warts, this is on the order of tens of watts. A 16 A circuit can deliver nearly 200 W at 12 V, and that could possibly replace a lot of wall-warts ... thing is still, if they are replaced by automotive-standard ones, where will the net savings be? Because all equipment has their own voltage and current ratings, I've got various wall-wart driven things here that range from 3 V DC to 30 V DC, and then there are 9 V AC, 24 V AC, and center-tapped 2x8 V AC (this last one is a brick on the cable) ... and no DC power distribution networks is going to be any improvement for these latter ones.
I once had a job at a place where they had common power supplies and low-voltage distribution to panels in the labs. The offerings included 5 V DC, 12 V DC, 120 V 3-phase 400 Hz AC, and a few others, but noise from equipment in one lab would appear in the power supplied to any of the others, so those low-voltage DC supplies turned out to be not really useful at all.
Some others have given some really good explanations on most of these questions, I'll try explaining a little more about the way the system deals with hardware.
The operating system kernel is this part of it that runs all the time and takes care of the scheduling of the various processes and of servicing the requests that these have for access to disks, keyboard, mouse, screen, serial ports, network ports, printers and so on. Since there is a lot of different varieties of these, the operating system uses a set of suitable drivers for each device. The devices and their associated drivers are of three major kinds: there are character devices, that work with streams of characters, such as the keyboard or the screen or the printer, and there are block devices such as disks, that deal with large chunks of data. The third kind of device are network interfaces that are not classified as traditional block or character devices at all.
The block and character devices are identified in the systems with their "major number". In older unix systems, there was two master tables of these device numbers, that indicated that for example character device number 1 would be a console, block device number 7 would be some kind of disk and so on.
On Linux systems, the same idea of major numbers are used, but the device driver programs usually request their own major numbers themselves, and the kernel builds these tables on boot, and we can see what this becomes through the pseudo-file /proc/devices.
Programs in the "user-space", that is all programs that we normally run as processes ourselves, have to access these devices through "node files", normally located in the /dev directory. When reading or writing to a device through these files, the operating system routes the requests for reading and writing to the appropriate driver since it knows the major number. Each driver provides a read() and a write() function to the kernel code that it calls for the device-driver to do its job. Both character and block devices may be read and written as if they were disk files, the block devices hold filesystems, and can additionally be mounted, and thus made accessible through a directory somewhere in the system.
The node files in /dev indicate a filename of a device, its major number, its minor number (this is used to tell several different devices of the same type apart), and of course whether this is a block or character device. On a long listing, (ls -l) the prefix shows as b or c for block or character device. Like other files, these may have read/write permissions set to control who may read or write the files. Most of them are owned by root. It is possible to create such node-files elsewhere in the file system, but that tends to be frowned upon.
In the above, very short, excerpt from /dev on this machine, we see the hard disk is block device major number 3, and the serial ports are character devices major number 4.
There is usually a whole slew of such node files under /dev, and they come in long series of names, beginning with some short prefix such as tty, pty, sda, hda, fd and so on. Generally, these prefixes follow a convention:
tty are serial lines, terminals, or consoles, (the abbreviation comes from "teletype"); pty are "pseudo-tty" devices, basically a kind of pipe that look much the same as a real tty to the programs reading and writing to them. Another character device of interest is the lp series, these are the parallell ports as used for printers.
fd ar
Such a DC-DC conversion already takes place inside the usual computer power supply. The first thing that happens to the mains power going in is that is being rectified, before a high-frequency oscillator circuit generates an AC voltage again, which then is rectified and produces the various voltages required by the server components.
As noted, low voltage DC implies fat and expensive conductors. So we could feed our servers with a higher DC voltage, and get away with reasonably sized conductors?
High voltage DC, while reducing the required current for the same power, still is rather nasty with its ability to sustain hot arcs. This makes for very expensive switches and fuses that won't just melt upon first attempt at disconnecting. I suspect one could use some variant of solid-state switches for breaking the load, combined with low rating switches to isolate the load electrically once the current is low, but this is the kind of complicated arrangement you'd expect to find in a power substation where there are several kilovolts of AC. Or maybe a rectifier station for a railway or streetcar system that runs on DC. But not the infrastructure of a server room...
I guess we could have individual UPS-supply combinations for each server machine, much like we have with batteries in notebook computers, but the fire-codes may require that all power can be switched off to the machines in the server room.
Then there's the economies of scale. Mains-powered equipment is readily available and have reasonably efficient power supplies. Distributing DC at the same power level (say a few kW) will require much more expensive components, whether these are sturdier switches or fatter conductors. Switching supplies are widely used and they are quite efficient, whether they have to convert AC to DC, DC to DC, or DC to AC, makes little difference; the energy savings and money savings will be in the cables and switches.
Pay for content. The revolution with the Web is that there is no limitations or anyone controlling the contents there. It used to be, with television, radio, and books, that only the select few producers were able to reach a large audience. Now this has changed to be determined by what you, YOU the reader and potential producer, have to say, and whether, or rather to what extent anyone's interested in it. Now anyone can read, and thanks to Google, anyone can find something they're looking for (as in it may not be what they want, but it will be what they need).
Had the web been created today by any media corporation or association of these, it would have been just another variation on the pay-for-content and "We produce, you consume" theme that is the bread and butter of the media companies today. They do not want to have any competition. And they do not want to surrender their control of the distribution channels.
That MS may restrict what can be run under their OS might be problematic enough, though possibly within their rights. Just look at their X-box. Of course, what will happen is that the developers will start using and advocating other platforms to an even larger degree.
And we can expect that MS forcing the production of Windows-only non-MS "PC" hardware (the Xbox is an MS loss-leader, and their baby) will go over like a lead balloon. Though I see the movements in that direction. Closed hardware is already endemic, all sorts of non-open specs apply to wireless network cards and screen cards. The extension to closing the rest of the machine is a reasonable extrapolation.
However, here is where someone will cry foul... what with MS being already a monopoly and using their operating system to entrench this? Yeah right, that is going to go really well with the courts, in either the US or Europe or the Far East.
And then consider also Massachusetts who just decided to go with all open-standard documents and open standard only -- this could already turn into a quite complicated issue for Microsoft, and if MS decide to make things difficult for the alternate operating systems, anything might happen, from them becoming just irrelevant for business use, to the re-opening of the monopoly case.
Even as easy as pre-installed Windows is to get running on a brand-new machine, I still get occasional phone calls from people whose newly purchased machine gives them trouble accessing the Internet, since the supplied modem cable was plugged into one of the sockets on the switch in the LAN...
Furthermore, the awareness of alternate operating systems or applications varies, many of them only know what Linux is based on what I have been telling them. These are not IT professionals, they see the computer as some kind of information-appliance, somewhat similar in function and purpose to some other well-known appliances such as typewriters, TV, filing cabinet and to some extent CD players. Like these, it can break (and they tend to be afraid of doing anything that could make that happen), and need to be fixed or replaced.
It is agurktid or "cucumber season" here in Norway. Whether this is from the tendency for the news bulletins to have about 3% substance, or it is from the period when all the papers had to write about was the size of the cucumbers on the town market... I don't know, but certainly, the first interpretation seemed to fit better in this case.
But some marketer probably came up with the change; even though these customers are a captive audience by the time they meet this "representative".
The starting premise, that manufacturers use the least expensive components they can get away with is no big news. This has been considered good engineering just about forever: use those parts that are good enough technically and least costly. The key is "good enough", though for what? The subsequent discussion on audio and video quality just indicates that the player is working as it should. As to distortion, there is only the subjective opinion that the cymbals don't sound as clear as they should, but there is no further indication of whether this comes from the player or the amplifier chain following.
Then the discussion proceeds to take apart and redo the power supply... Not components in the signal path, but the power supply. A switching power supply that is powered direct from the mains, with X and/or Y-rated capacitors and inductors in the power entrance, and a somewhat carefully orchestrated feedback loop that causes its components to oscillate and generate the correct DC voltages for the electronics. Apparently, (and fortunately) most of the really important components are surface-mount and thus not amenable to this kind of tinkering. The argument goes on that this more "stiff" power is needed for more accurate bass-response. Hello? this isn't a 30 Watt tube-amplifier we are talking about here, where such an argument might hold, but something that puts out a few tens of milliwatts of power into maybe 600 Ohms at the most. Unless the power supply is pathologically feeble, this is really just -- marketing-speak, to use a polite term.
However, even if putting in larger filter electrolytic capacitors might be harmless, the replacement of the X-rated capacitor with an "Auricap" which is evidently NOT X-rated, sounds dodgy, as in potential fire hazard. The Auricaps seem to be marketed as non-electrolytic replacements for electrolytic capacitors in the signal path, and might do a fine job here, but we're not expecting any of our precious signal to enter the mains are we?
The fact that there is an X-rated capacitor there at all and not just a cheaper one, is that this is sitting across the mains voltage, and has to conform to specific requirements from the UL, CSA, TUV and so on, lest they refuse to List or certify the equipment for sale. And probably more important, that the component fails safe and does not start a fire.
Googling for "aurocaps" shows several sites catering to the same merry lot who depends on "oxygen-free" cables, and praise the virtues of the gold-plated RCA connectors... There is a reason why professional kit uses XLR or BNC connectors, or even 1/4inch jacks by the way, and it has to do with mechanical and electrical stability and shielding, not any magic properties of gold or nickel that makes one good and the other bad.
It goes on about replacing more of the bypass capacitors in the digital processing section, and mentions the possibility of clock jitter. Technically correct. But no quantitative information, no measurements done on a distortion analyzer or even a picture of the signal on an oscilloscope. Just all this non-scientific hand-waving that if we put in more expensive capacitors the sound will be better.
Finally, op-amps and possible replacements. Again, the observation is that the amplifiers are low cost, and obviously we could put in better and more expensive ones. Low cost is not the same as crummy; had the manufacturer put in really bad ones, everyone would have heard. Again, it is a matter of good enough, though the only parameter that might make a difference would be the noise of the amplifier. Unless they are really atrocious (with obvious effects on sales), gain and slew-rate would not matter, except for marketing purposes.
My guess for what might constitute the perceived "improvement" in this case, is that the frequency response of the audio chain has changed, boosting the higher frequencies, and thus made the modified unit appear to sound better.
Most of the body parts like quarter panels and door bodies, floor, hood, etc. would probably best be made out of sheet metal, suitably bent or pressed. Other structural parts probably could be best made out of pipe stock. There still exist drawings of these components, and it wouldn't be impossible for some metal shop to make them, even considering adjustments for the different mechanical parts.
But smaller, irregularly-shaped parts, such as door handles, instrument panel, light-fixtures and much of the cast chromed parts, could very well be made with some of these new methods.
The resulting vehicle would perhaps become a '07 Bel Air? :)
Another thing is that these fabrication methods can be used for making new specialty engine or other parts for old cars in general, where no modern useable equivalent is available.