Of 3000 random users many or most of them will be on broadband. Ample bandwidth for serving live MP3s to multiple people.
My experience of using other P2P-apps tells me otherwise. But then again, you could be right.
However of 3000 random users I'm sure you will be able to find plenty with various kinds of non-top-40 collections. Doing an artist or genre search will probably pull up general matches, and at that point there's fair chance their other songs are similar.
Most typical p2p-users will have large more or less unsorted collections containing Weird-Al, Donald Duck orgasm, top40 stuff, and then some music they actually like. Grabbing 10 random songs from that is not going to produce a reasonable playlist.
Genre-search has always been useless on p2p-apps. That is because the genre that is in the id3-field of mp3 files is (1) too limited in selection (only a fixed number of genres), and (2) wrong (set by clueless user), and therefore typically useless.
A reasonable genre-search would be something like "what DJ foo would play on the show quux". Not something like "pop", "rock", "alternative", "classical", or "jazz".
So basically we can now choose between 3000 random users random 10-song playlists streamed over inadequate bandwidth without the ability to find any songs beyond the typical top-40 songs, or to save them. Add to that weird claims of legality, and privacy concerns from the scanning of the harddrive, and it suddenly doesn't sound so nice anymore. Not that it did sound any nicer in the first place. Most p2p apps already suck, making it even more artificially restricted doesn't really help.
In theory this i true, but in reality it's impossible to create a one-time-pad.
Huh? You can't have tried hard. Try throwing a dice enough times, and record the results. That is your one-time pad.
You'd need quantum-mechanics to create a real one-time-pad, because we don't have the possiblity to create such randomness with a computer.
A dice is a simple and effective way of getting random numbers. I'm unsure about what you mean by "need quantum mechanics". Apart from being needed simply because it's another part of the laws of physics, you are obviously mistaken.
And while old cpus doesn't include a hardware random number generator (which isn't really that hard to build), new ones from AMD and VIA does. And even on older computers, you could always add a genuine random number generator as an add-on connected to some bus.
But even if you don't have a hardware random number generator, it is nowhere as hard to create random numbers as you suggest. That is because real computers (as opposed to Turing machines) do input and output, and when you combine enough different inputs, what you get is for all practical purposes random./dev/random on linux is just one example.
To create real randomness, a PRNG has to have 3 properties:
Uh, no! Whatever constitutes real randomness is more of a philosophical question. But a pseudo-random number generator is certainly not real randomness. That is why the P (for pseudo) is there.
I have my computer inside a faraday cage, inside a faraday cage, inside a faraday cage, in a locked and alarmed hidden room in the basement. It is kept together with five other dummy computers of exactly the same type, running various random computer-intensive tasks, as well as various other equipment generating electronic and acoustic noise, such as TVs, various engines, etc... Everything is powered by a diesel generator coupled with an UPS. Instead of a hard-drive I use a CD-ROM for the system software and a ramdrive for/home.
If someone enters the room (or any of the consecutive faraday cages) without typing in the correct 15-digit code at the alarm panel within 2 seconds, the power automatically shuts off, loosing any data on the ramdrive. It also detonates a homemade bomb of about 100 kg explosives, 500 kg rusty nails and broken glass, and about a ton of sulphuric acid, and causes the five samples of smallpox virus stored outside the house, to be released.
At least now I know my hiscore-data for wolfenstein 3d is secure!
The following steps should'nt be necessary to do more than once per unemployment period. Preferably as early as possible, as it is the most guaranteed way of finding a job you actually like!
Find companies that you know offer jobs you might be interested in
Write a general application form telling them what kind of jobs you might be interested in, and why, as well as a copy of your CV
Furthermore, these steps should be followed at least once a week:
Find ads for jobs you are interested in (in newspapers, internet job database, unemployment agency, etc...)
Read them
If there are some of them you are interested, think you have a chance to get, and haven't already applied, apply
Concurrently, the following steps must be followed every day
Check mail (yes, even snail-mail!), be available on phone (or at least check your answering machine)
If you are asked to come for an interview, say yes, take a note of the company, address, contact persion, time and date
If you have an interview today, take a shower, dress up somewhat responsibly, and try to be there on time
Finally, a tip regarding CV's. It should be honest, to-the-point, and keep mostly only informaton of interest to the potential employer. E.g. If you are applying for a bodyguard job, make sure to cover all your military training in detail, but only briefly mention you PhD's. If you are applying for a job as a rocket scientist, you can do the opposite. If there are holes in your CV, such as periods of sickness, depression, etc, be brutally honest, and explain why.
To keep yourself organized, you need a pen, and two sheets of paper, one with the names of companies you have already applied for, and another with company/address/contact person/date/time/ for interviews you have agreed to meet up for.
If these steps are too complicated, and you feel you need a DBMS to keep track of you job-hunt, you are doing something wrong.
Huh? Alsa? What's wrong with alsa? It's a sound-driver. It makes it possible to get input and output to and from your soundcard(s). It supports hardware and software mixing. And with linux/alsa/jack you generally get lower latency and higher flexibility than with windows. It also supports several professional soundcards...
Now audacity, I can agree. It's not in any way comparable to many professional audio tools for windows. But hey, if you wan't professional audio tools, why pick on audacity, it doesn't look like it was intended to be one. Now, if you had picked on wired, muse, ardour, rosegarden, ecasound, csound, pd, jmax, cmt, rtcmix , nyquist, sweep, jamin, or any of the other more "professional" tools, I could have understood you. But picking on audacity for not being suitable for professionals is like picking on soundrec.exe for much the same reason.
Yeah, dasher looks pretty cool. I wonder if they used dasher to write the homepage, where it says they also have a "Japapese" version available (emphasis mine, typo them...).
ML: if objThisControl <> objTheControlICareAbout then.. else...
Forth: objThisControl objTheControlICareAbout <> if... then
PostScript: objThisControl objTheControlIcareAbout ne {...} if
It seems to me it's programming you don't like, not BASIC.
PS: There are plenty of reasons to hate BASIC. This is not one of them.
PPS: The PostScript example probably does a deep comparison. It's been a long while since I touched PostScript. Standard ML doesn't have a meaningfull way of comparing references.
Well, in a manner of speaking the human species is open-source.
Yeah, in the same manner of speaking that Microsoft Word is open source. Let's face it, raw DNA is not the preferred way to modify the human genome. Neither is a binary executable the preferred way to modify a word-processor.
It's just that reading the source (i.e. the DNA) is something we haven't quite mastered yet.
Ok, that's a possibility too. But I can't really see any plausible reason for nature to make it easy for us to decode DNA. It's meant for execution by cells, not for easy comprehension (or modification) by genetic engineers.
And just like open-source software the human being is constantly being developed upon, albeit so very slowly that we cannot see it for ourselves. But each new baby that's born (the nightlies, hahaha) is a little different than the previous generation. Sometimes this backfires on the whole community *cough*georgewbush*cough* but in general there's improvement.
Actually, there's no good reason to believe human evolution is for the better. Ever since humans discovered farming, we no longer need to be intelligent enough to outwit our prey, or strong enough to kill it, or fast enough to catch it, or having a good enough immune system to eat rotten food, or being able to care for our children in this environment for umpteen years, etc...
In fact, in modern society it's even worse. Due to the fact that the state will take care of you and your children even if you can't do it yourself, the only skill you need to create surviving offspring is to have low enough demands to get laid by someone (no matter how stupid, annoying, ugly, etc), and to be stupid enough not to wear condoms.
Humans have probably "devolved" ever since the neanderthals, and as the society gets "better", the individuals detoriate at an ever increasing rate.
An interesting side-note (just to be politically correct), is that those who advocate racial purity (e.g. nazis), seems to prefer those who have lived in modern society for the longest period of time. They probably should prefer bushmen instead.
Of over 30 billion e-mails being sent daily, some experts estimate that over 40% is Spam.
Huh? Only 40%?
What is people doing these days? Doing nothing but sending emails? What happened to talking face-to-face, or phones? I am certainly getting a lot more than 40% spam. 95% would be more believable.
There are two possible ways to store the bytes of a 16 bit word: least significant byte first (called little endian), and most significant byte first (called big endian).
Actually, there are a lot of other ways. Let's start with the most significant bit. We can place it at pos 1..16. Now there are 15 possible positions for the next bit, and so on... So we end up with 16! = 20922789888000 different arrangements
That is of course, if we limit ourselves to simply rearranging bits. Of course we can do other tricks as well, such as noting them, xoring them, or other more complex functions such as add, subtract, multiply, divide, etc. The absolute limit on representations is found when we consider all 1-1 functions from 1..2^16 to 1..2^16. That is: (2^16)!, a mindboggingly high number that I won't even bother to calculate.
On the other hand, for real hardware, you are probably right. But if you consider 32-bit numbers, there do exist architectures that are not just 'unix' or 'xinu', but also stuff like 'nuxi'.
Yes, it's a "classic", in the sense that it's been there for a while, many good programmers have read it, and many good programmers recommend it. Disclaimer: I'm not among those who have read it, or recommended it.
I found it very funny. If gentoo existed while I was in the "larval stage" of using linux, I would surely be a gentoo-ricer myself. Instead, I was mostly a slackware-ricer, with two or three extra partitions to try out a new linux distro a month.
Then I discovered debian, and since it was the only system I could easily keep up-to-date (let's face it, in those days, most distros didn't easily upgrade), it kind of stayed on my HD. I don't know how many years passed (5-10?), and I'm still using debian, and I still haven't reinstalled (except once, when I replaced my old computer). I've come to the point where I don't want to waste my time using another distro, as debian "testing" is good enough for me.
I've never even tried gentoo, but I certainly recognize myself in the attitudes displayed there. Go on kids! Use gentoo for a while, then when you become bored of being ricers, try debian on a separate partition. If you have a broadband-connection, it will probably stick there. I honestly think it's the best path...
Eh.. Did you care to investigate the matter yourself? The authors got it right. Chernobyl was bad, Bhopal was worse. People tend to have an irrational fear of anything "nuclear", but are happy to wipe their ass with dioxins every day.
ALICE: Tell me about your likes and dislikes?
Human: i like almost everything, except boredom
ALICE: I'm not sure if I like it. I really like ELIZA. I like electricity.
Human: do you know eliza personally?
ALICE: No, tell me about him, her or it.
I have no trouble accepting that you can coax your CASE tool into drawing something that looks like a DFD, I can't see how you can preserve it's semantics in UML. I can probably coax my CASE tool to write modernist art too, but it has absolutely no meaning in UML (not that modernist art is particulary meaningful for anyone, but that's besides the point).
Since activity diagrams are basically flowcharts, and flowcharts and DFD's are totally different things, I can't see how activity diagrams can be DFDs too, no matter how much you "stereotype". That was my question.
Interesting. Last time I heard, on some UML course I was forced to participate in, at my last job, the hired expert said that the UML gang didn't like dataflow diagrams, because of some ideas they had about use of model checking (use of formal methods) in the case tool, or something like that, which basically meant that DFD would break UML somewhat. Not that I've really investigated this issue myself, as I'm happy to draw DFD's on napkins anyway, and never really saw the point of having one unified notation for every OO design issue, unless that notation is the same thing you would use to implement the darn thing in (i.e. C++, java).
But if they are now at least somewhat supported, that is a good thing.
But despite your post, I fail to see how it can be done, as a dataflow diagram has completely different semantics from an activity diagram, and just loading some fancy graphics to replace the fancy graphics in an activity diagram won't change that. If you can post a link to an example, I would be very happy. Preferably of the napkin variety, as you shouldn't need a $200 windows-only case tool, just to mock up a dataflow diagram for a simple control system.
This goal is often achieved with a Data Flow Diagram ...[snip]...
Eventually, we got to UML from there after 10-15 years of haggling and stumbling; some argue that it still isn't working out so well.
Yeah, well. Some of us are complaining about the lack of dataflow diagrams in UML;-) If it doesn't even support that, what do we need a standardized notation for? Drawing stick-figure men?
There is no way to use exceptions that is generally readable. And rampant uncontrollable gotos, which is a significant aspect of exceptions isn't avoidable (please enlighten me otherwise).
Yes it is. That is exactly why we have exceptions, to avoid thousands of if-tests to test for every particular error everywhere. If you e.g. make a simple recursive descent parser, it makes sense for the parser to throw exceptions instead of infecting all your code with tests for errors. You are generally not in a position to "fix" the error anyway, all you want is to say which part of the input is wrong, and that can be done at a single place in your code.
Without exceptions, you must either write each routine to be able to return a new error code, and make each routine test for that all the places it calls other routines. This is dead ugly. Alternatively, simply longjmp out of the entire block, but that is the same as with exceptions, except that it is uncontrollable.
And whether libraries generate exceptions instead of return codes isn't typically under your control. It's under the control of the creator of the library.
Yes, that is the way the world is. Writing general purpose libraries is always hard. And unfortunately, not every library writer has a clue. If you are really offended by the braindead choices done in some badly designed library (e.g. CORBAs C++-bindings) that you have to use, there's always the possibility of wrapping up the stupid code.
Any exception handler is going to be a nested block of code. and if you allow an outer exception handler to be reached you really have a totally unpredictable result as far as the curent state of the machine. You have no choice but to write ugly code if exceptions are around.
No, it is not uncontrollable. That is exactly why C++ offers the "resource acquisition is initialization" idiom, java and C# offers try-finally blocks, common lisp offers unwind-protect, etc... This is great stuff, it gives you the opportunity to jump out of the offending code quickly and painlessly, while at the same time being guaranteed cleanup of resources and state in a predictable manner.
And in most cases, you don't even need them. Unless your code goes about allocating resources or modifying global state, you could care less about what happens with local variables on the stack you are jumping out of.
Add to that C#'s interminable nesting of classes generally, and you have code moving several levels deep to the right.
There's no way to make such code readable.
Ever heard of abstraction? If you, or your co-workers isn't capable of writing short readable methods that spans no more than a screenfull, with a nesting-level below 3, it is hardly the languages fault.
I can see where you are coming from. I've seen some horrible code using exceptions in obtuse ways. If you are always testing for errors immediately anyway, exceptions doesn't make your code any more readable (in fact, it makes it worse, which we both seem to agree upon).
The power of exceptions comes from being able to write code that is clean, safe, and isn't threaded with error-checking everywhere. You can start writing modular code again: normal program logic in one place, error-reporting somewhere else. And the system will take care of the details, if you learn to use it properly.
Using exceptions isn't a device to stop idiot programmers from having to think about error conditions. Like any other way of handling errors, it requires you to think and plan your way of attack. What it does give you, is the tools to write cleaner and more modular code, if you know how to use them.
Slashdot Mirror
My experience of using other P2P-apps tells me otherwise. But then again, you could be right.
However of 3000 random users I'm sure you will be able to find plenty with various kinds of non-top-40 collections. Doing an artist or genre search will probably pull up general matches, and at that point there's fair chance their other songs are similar.
Most typical p2p-users will have large more or less unsorted collections containing Weird-Al, Donald Duck orgasm, top40 stuff, and then some music they actually like. Grabbing 10 random songs from that is not going to produce a reasonable playlist.
Genre-search has always been useless on p2p-apps. That is because the genre that is in the id3-field of mp3 files is (1) too limited in selection (only a fixed number of genres), and (2) wrong (set by clueless user), and therefore typically useless.
A reasonable genre-search would be something like "what DJ foo would play on the show quux". Not something like "pop", "rock", "alternative", "classical", or "jazz".
So basically we can now choose between 3000 random users random 10-song playlists streamed over inadequate bandwidth without the ability to find any songs beyond the typical top-40 songs, or to save them. Add to that weird claims of legality, and privacy concerns from the scanning of the harddrive, and it suddenly doesn't sound so nice anymore. Not that it did sound any nicer in the first place. Most p2p apps already suck, making it even more artificially restricted doesn't really help.
Huh? You can't have tried hard. Try throwing a dice enough times, and record the results. That is your one-time pad.
You'd need quantum-mechanics to create a real one-time-pad, because we don't have the possiblity to create such randomness with a computer.
A dice is a simple and effective way of getting random numbers. I'm unsure about what you mean by "need quantum mechanics". Apart from being needed simply because it's another part of the laws of physics, you are obviously mistaken.
And while old cpus doesn't include a hardware random number generator (which isn't really that hard to build), new ones from AMD and VIA does. And even on older computers, you could always add a genuine random number generator as an add-on connected to some bus.
But even if you don't have a hardware random number generator, it is nowhere as hard to create random numbers as you suggest. That is because real computers (as opposed to Turing machines) do input and output, and when you combine enough different inputs, what you get is for all practical purposes random. /dev/random on linux is just one example.
To create real randomness, a PRNG has to have 3 properties:
Uh, no! Whatever constitutes real randomness is more of a philosophical question. But a pseudo-random number generator is certainly not real randomness. That is why the P (for pseudo) is there.
If someone enters the room (or any of the consecutive faraday cages) without typing in the correct 15-digit code at the alarm panel within 2 seconds, the power automatically shuts off, loosing any data on the ramdrive. It also detonates a homemade bomb of about 100 kg explosives, 500 kg rusty nails and broken glass, and about a ton of sulphuric acid, and causes the five samples of smallpox virus stored outside the house, to be released.
At least now I know my hiscore-data for wolfenstein 3d is secure!
The following steps should'nt be necessary to do more than once per unemployment period. Preferably as early as possible, as it is the most guaranteed way of finding a job you actually like!
Furthermore, these steps should be followed at least once a week:
Concurrently, the following steps must be followed every day
Finally, a tip regarding CV's. It should be honest, to-the-point, and keep mostly only informaton of interest to the potential employer. E.g. If you are applying for a bodyguard job, make sure to cover all your military training in detail, but only briefly mention you PhD's. If you are applying for a job as a rocket scientist, you can do the opposite. If there are holes in your CV, such as periods of sickness, depression, etc, be brutally honest, and explain why.
To keep yourself organized, you need a pen, and two sheets of paper, one with the names of companies you have already applied for, and another with company/address/contact person/date/time/ for interviews you have agreed to meet up for.
If these steps are too complicated, and you feel you need a DBMS to keep track of you job-hunt, you are doing something wrong.
Now audacity, I can agree. It's not in any way comparable to many professional audio tools for windows. But hey, if you wan't professional audio tools, why pick on audacity, it doesn't look like it was intended to be one. Now, if you had picked on wired, muse, ardour, rosegarden, ecasound, csound, pd, jmax, cmt, rtcmix , nyquist, sweep, jamin, or any of the other more "professional" tools, I could have understood you. But picking on audacity for not being suitable for professionals is like picking on soundrec.exe for much the same reason.
Nope. It's called endianness.
Yeah, dasher looks pretty cool. I wonder if they used dasher to write the homepage, where it says they also have a "Japapese" version available (emphasis mine, typo them...).
Just looking at constructs like "If Not (objThisControl Is objTheControlICareAbout) Then" makes my stomach churn..
Yeah, that really makes BASIC look bad, doesn't it?
Let's see:
C, java, etc: if (objThisControl != objTheControlICareAbout) ...
lisp: (unless (eq ObjThisControl objTheControlICareAbout) ...)
ML: if objThisControl <> objTheControlICareAbout then .. else ...
Forth: objThisControl objTheControlICareAbout <> if ... then
PostScript: objThisControl objTheControlIcareAbout ne {...} if
It seems to me it's programming you don't like, not BASIC.
PS: There are plenty of reasons to hate BASIC. This is not one of them.
PPS: The PostScript example probably does a deep comparison. It's been a long while since I touched PostScript. Standard ML doesn't have a meaningfull way of comparing references.
Yeah, in the same manner of speaking that Microsoft Word is open source. Let's face it, raw DNA is not the preferred way to modify the human genome. Neither is a binary executable the preferred way to modify a word-processor.
It's just that reading the source (i.e. the DNA) is something we haven't quite mastered yet.
Ok, that's a possibility too. But I can't really see any plausible reason for nature to make it easy for us to decode DNA. It's meant for execution by cells, not for easy comprehension (or modification) by genetic engineers.
And just like open-source software the human being is constantly being developed upon, albeit so very slowly that we cannot see it for ourselves. But each new baby that's born (the nightlies, hahaha) is a little different than the previous generation. Sometimes this backfires on the whole community *cough*georgewbush*cough* but in general there's improvement.
Actually, there's no good reason to believe human evolution is for the better. Ever since humans discovered farming, we no longer need to be intelligent enough to outwit our prey, or strong enough to kill it, or fast enough to catch it, or having a good enough immune system to eat rotten food, or being able to care for our children in this environment for umpteen years, etc...
In fact, in modern society it's even worse. Due to the fact that the state will take care of you and your children even if you can't do it yourself, the only skill you need to create surviving offspring is to have low enough demands to get laid by someone (no matter how stupid, annoying, ugly, etc), and to be stupid enough not to wear condoms.
Humans have probably "devolved" ever since the neanderthals, and as the society gets "better", the individuals detoriate at an ever increasing rate.
An interesting side-note (just to be politically correct), is that those who advocate racial purity (e.g. nazis), seems to prefer those who have lived in modern society for the longest period of time. They probably should prefer bushmen instead.
Huh? Only 40%?
What is people doing these days? Doing nothing but sending emails? What happened to talking face-to-face, or phones? I am certainly getting a lot more than 40% spam. 95% would be more believable.
Actually, there are a lot of other ways. Let's start with the most significant bit. We can place it at pos 1..16. Now there are 15 possible positions for the next bit, and so on... So we end up with 16! = 20922789888000 different arrangements
That is of course, if we limit ourselves to simply rearranging bits. Of course we can do other tricks as well, such as noting them, xoring them, or other more complex functions such as add, subtract, multiply, divide, etc. The absolute limit on representations is found when we consider all 1-1 functions from 1..2^16 to 1..2^16. That is: (2^16)!, a mindboggingly high number that I won't even bother to calculate.
On the other hand, for real hardware, you are probably right. But if you consider 32-bit numbers, there do exist architectures that are not just 'unix' or 'xinu', but also stuff like 'nuxi'.
Yes, it's a "classic", in the sense that it's been there for a while, many good programmers have read it, and many good programmers recommend it. Disclaimer: I'm not among those who have read it, or recommended it.
Then I discovered debian, and since it was the only system I could easily keep up-to-date (let's face it, in those days, most distros didn't easily upgrade), it kind of stayed on my HD. I don't know how many years passed (5-10?), and I'm still using debian, and I still haven't reinstalled (except once, when I replaced my old computer). I've come to the point where I don't want to waste my time using another distro, as debian "testing" is good enough for me.
I've never even tried gentoo, but I certainly recognize myself in the attitudes displayed there. Go on kids! Use gentoo for a while, then when you become bored of being ricers, try debian on a separate partition. If you have a broadband-connection, it will probably stick there. I honestly think it's the best path...
Eh.. Did you care to investigate the matter yourself? The authors got it right. Chernobyl was bad, Bhopal was worse. People tend to have an irrational fear of anything "nuclear", but are happy to wipe their ass with dioxins every day.
Wow. Are you for real?
ALICE: Tell me about your likes and dislikes?
Human: i like almost everything, except boredom
ALICE: I'm not sure if I like it. I really like ELIZA. I like electricity.
Human: do you know eliza personally?
ALICE: No, tell me about him, her or it.
Most likely that would help, yes.
Buy a cheap no-brand CD-player with a large PLAY and EJECT button. The other buttons are there mostly for geeks.
Why would it be a good thing? Someone in particular you want sent there?
Since activity diagrams are basically flowcharts, and flowcharts and DFD's are totally different things, I can't see how activity diagrams can be DFDs too, no matter how much you "stereotype". That was my question.
But if they are now at least somewhat supported, that is a good thing.
But despite your post, I fail to see how it can be done, as a dataflow diagram has completely different semantics from an activity diagram, and just loading some fancy graphics to replace the fancy graphics in an activity diagram won't change that. If you can post a link to an example, I would be very happy. Preferably of the napkin variety, as you shouldn't need a $200 windows-only case tool, just to mock up a dataflow diagram for a simple control system.
You are probably being serious. Oh, well...
Yeah, well. Some of us are complaining about the lack of dataflow diagrams in UML ;-) If it doesn't even support that, what do we need a standardized notation for? Drawing stick-figure men?
Yes it is. That is exactly why we have exceptions, to avoid thousands of if-tests to test for every particular error everywhere. If you e.g. make a simple recursive descent parser, it makes sense for the parser to throw exceptions instead of infecting all your code with tests for errors. You are generally not in a position to "fix" the error anyway, all you want is to say which part of the input is wrong, and that can be done at a single place in your code.
Without exceptions, you must either write each routine to be able to return a new error code, and make each routine test for that all the places it calls other routines. This is dead ugly. Alternatively, simply longjmp out of the entire block, but that is the same as with exceptions, except that it is uncontrollable.
And whether libraries generate exceptions instead of return codes isn't typically under your control. It's under the control of the creator of the library.
Yes, that is the way the world is. Writing general purpose libraries is always hard. And unfortunately, not every library writer has a clue. If you are really offended by the braindead choices done in some badly designed library (e.g. CORBAs C++-bindings) that you have to use, there's always the possibility of wrapping up the stupid code.
Any exception handler is going to be a nested block of code. and if you allow an outer exception handler to be reached you really have a totally unpredictable result as far as the curent state of the machine. You have no choice but to write ugly code if exceptions are around.
No, it is not uncontrollable. That is exactly why C++ offers the "resource acquisition is initialization" idiom, java and C# offers try-finally blocks, common lisp offers unwind-protect, etc... This is great stuff, it gives you the opportunity to jump out of the offending code quickly and painlessly, while at the same time being guaranteed cleanup of resources and state in a predictable manner.
And in most cases, you don't even need them. Unless your code goes about allocating resources or modifying global state, you could care less about what happens with local variables on the stack you are jumping out of.
Add to that C#'s interminable nesting of classes generally, and you have code moving several levels deep to the right. There's no way to make such code readable.
Ever heard of abstraction? If you, or your co-workers isn't capable of writing short readable methods that spans no more than a screenfull, with a nesting-level below 3, it is hardly the languages fault.
I can see where you are coming from. I've seen some horrible code using exceptions in obtuse ways. If you are always testing for errors immediately anyway, exceptions doesn't make your code any more readable (in fact, it makes it worse, which we both seem to agree upon).
The power of exceptions comes from being able to write code that is clean, safe, and isn't threaded with error-checking everywhere. You can start writing modular code again: normal program logic in one place, error-reporting somewhere else. And the system will take care of the details, if you learn to use it properly.
Using exceptions isn't a device to stop idiot programmers from having to think about error conditions. Like any other way of handling errors, it requires you to think and plan your way of attack. What it does give you, is the tools to write cleaner and more modular code, if you know how to use them.