But aren't there books, written on paper, describing RLE? Maybe the answer is to store some books describing file formats in an inert atmosphere?
The problem is not to decode RLE files. The problem is to have the idea to interpret them as RLE. Look how the fax machine compresses the data in RLE: It has two colors, white and black. It always assumes, the first color is white and then sends a byte with the number of white dots to follow. Then it sends a byte with the now following number of black dots and repeats this again and again. If for some reason the first dot is black, it starts with a 0-byte (No white dots), then the number of black dots to follow. If there are more than 255 dots of one color following each other, then it sends a FF-byte, then a 0-byte, and then the remaining dots in the next byte.
It is for itself a quite easy algorithm, and it is easy to encode and to decode (fax machines should be reliable and simple).
But imagine you get a file with all the bytes in it. How do you tell it's RLE data encoding a picture? Maybe it's MIDI data instead? Or the index of a filesystem?
The rosetta stone contained the message in Ancient Greek (a dead but widely known language at the time of deciphration), Coptian and Hieroglyphic. Even though Coptian was at least known to some specialists and people able to read Ancient Greek were abundant at the time (and still are), it took about 25years to decipher the hieroglyphic texts.
And this was with a language which itself was very easy mapped to the letters (every consonant mapped to a letter, vowels omitted).
The rules which encode a file may be much more complicated. Look just at the most common compression methods (Run Length for instance), how they just add another layer above the already encoded contents. And they remove something very important for deciphration, the redundancy, out of the data. Then the subjects that are stored in files are much more diverse. We have not only language, we have music and graphics, 3D data and cryptographic certificates, configuration files and program binaries.
Just to be able to know what the file is about and thus have an idea how to get started can prove to be more complicated than any deciphration from archaeologic texts.
That's what I call "the wrong way around". You should have it really stored at the same place on every installation. But to the user you should have a localisation layer, that translates "Program Files" to whatever her locale wants to have. So no program has ever to take care where to store its files, and the localisation layer transforms it to whatever the local user wants.
What? You are NOT using the german version? I hate it when programs never check for the default install path and just hardcoded assume, it would be C:\Program Files\. And I am pretty sure, it would be something like C:\Fiches des Programmes\ in the french version.
Don't tell me, Windows got it right from the start. They messed it up like everyone else. And this in the same company.
A paper ballot and a pen is the only form of ballot I trust. And if they don't count the ballots AT THE POLLING PLACE in plain view of the public BEFORE they ship them off to the court house you can't trust the result.
Are you trying to tell me, that in the U.S. the votes don't get counted in the polling place for each voting district in the open public?
Silly me to think it would be a prerequisite for a democratic election to have a local public control how the votes are counted. But on the other hand: We surely have to learn from the U.S. how to screw up a simple thing like an election by trying to automatise everything.
Yes I know you can also screw up while trying to vote with paper and pen. That happens. If it happens to you just write on the paper "INVALID VOTE" after miscasting your vote or if you think your vote could be missinterpreted because you didn't know how to make a cross at the right place. At least you know it's only you who screwed up your vote.
You only have to release source code if you -distribute- your modified kernel.
To be more specific: You have only to release the source code to the people you are distributing the binaries to. So if someone modifies the Linux kernel and gives the binary kernel to someone else, he has to make available the source code of the modified kernel in the same way to exactly the same person.
What you have to permit the recipient of both binary and source code is to do the same, modify the source, create new binaries and give them away to someone else together with the newly modified source code.
Those other ones are not necessarily the common public, they are just the people you are giving your modifications to. So GPL doesn't require you to publish your source. It requires you to make your changes in the source available in the same way you make the modified binaries available. If the target audience is just a small group of people, then the modified source is only available to the same small group of people. If you using the modifications inhouse, the modified source can stay inhouse.
Spindler spots jurisdictional uncertainties for all parties involved: Developers may be held liable if software does not work as expected, even if they only participated marginally in the development, rather than being a lead developer. Employers could walk on thin ice if they pay employees for writing Open Source Software. And buyers of such software must be prepared that liability is limited to the criteria common for items given away for free, i.e. severe negligence only.
But the same is valid for all commercial software in Germany too. EULAs have repeatedly been denied validity because of the german contract law. EULAs are a contract between you and the author of the software. But because you didn't buy the software directly from the author, but from a third party (the reseller, the company bundling the software with a computer etc.pp.), EULAs can't be enforced. All the author can impose on you is the priviledges he gains from the Author's Right (Urheberrecht).
This makes software under GPL in no way different than any commercial software you buy in Germany from a liability point of view.
What you were linking to is the Swisstool, not the Swiss Army Knife. Rather look at this page for reference. Ok, I have to admit that yours look more like a Leatherman Tool after all;)
It is not. I did actually read the presentation of the intervention given in the patent statement, and I remember some parts of the presentation we once got.
Here are the main differences to Akamai and Squid:
1. Mirror Image determines which source to use by looking where the request comes from first. This is actually done by having ".mirror." (or another subdomain similar to this one) as part of the requested URL and the subdomain's resolver being part of the Mirror Image system. So the resolver knows where the request comes from and thus triggers the right source to actually deliver the requested information.
2. Squid and Akamai always deliver the content themselves, either from cache or by first requesting the information from the server and delivering it then to the user. Mirror Image instead, if not able to serve the request from cache, sends a 302 Redirect page to the browser (so the browser sends a new request to the original source) and then itself requests the information and stores it in the cache. Thus, Mirror Image claims, the responsivity of the cache increases.
So Squid and Akamai don't have anything to worry about because of this patent. They are operating a different way not covered by the patent.
Having worked with Mirror Image I have to say that the way Mirror Image is doing the caching differs strongly from Akamai's.
While Akamai is putting cache servers in many IP provider's locations (I think more than 5000 so far), Mirror Image is concentrating its caches in about 20 locations connected to the big exchange and peering points. The Mirror Image presenters were explicitely stressing this point and that this other approach is the key to Mirror Image's success. So I guess the patent covers the Mirror Image Way Of Doing Things rather than the idea to cache websites to speed up transfer rates.
You're not disputing the fact that AT&T owned System V at one point in time, are you?
No, he's just disputing AT&T's right to sell (or give away) the termination clause, because this is a mutual agreement and thus only changeable if IBM agrees (as stated clearly in section 4). If AT&T could have assigned the right to any other entity it would have been "AT&T or assignee" in the agreement and not just "AT&T".
It's a lot like the AIDS/HIV test. You want every potential positive test to show up so you can follow it up but you do not want to miss a potentially infected person.
Normally every medical test in fact consists of two tests. The first one is called the efficient test, the second one the effective test.
The first one is to throw out as much uninteresting people (the ones without any sign of sickness) as possible without missing any of the interesting people (the ones that are sick). In Computer Science you call a test like this "Trivial Reject". It should be fast and cheap to save on costs (of every type, like monetary, computational time...).
The second test is supposed to catch as much of the interesting people (sick ones) as possible while throwing out pretty surely all of those who are in fact healthy and thus uninteresting for the matters of the test. This leaves you with a test population which consists of almost all sick people and some very few false positives. Because of the shrunken number of persons this test can be more elaborate and expensive.
The problem with all those tests is, that with the low frequency of the usual dangerous plague you are testing for, the false positives (even though they may be very small compared to the whole of the population) still contains more people than the actual sick ones. I remember some exercises where we had to calculate the risk of actual being sick after a positive test for given frequencies of occurance of the plague in the population and the characteristics of the test. Often the result was that you have a 99% chance of being a false positive even though the test itself was pretty good.
Same is valid for the frequency of the actual terrorists in the population of all U.S. domestic flyers. While there were 19 terrorists in the 9/11 attack, there have been billions of individual flights in the recent years. If your tests are being 99.999% sure in clearing a innocent flyer, it would still mean that this test applied to one billion flight passengers gives you 10 000 false positives.
(This number is one of the reasons why the Romans once decided for "in dubio pro reo". Because the number of lawful citizens is quite high compared to the number of criminals, the average screening of larger populations results in more false positives than actually caught thieves. So you have always to prove individually that someone is guilty beyond any doubt before you can actually call them guilty.)
In this special case the screening for names itself is a very inaccurate test. There are names in some countries being too prevalent to distinguish between people. For Corea it may be the name 'Kim', for India 'Singh', for Russia 'Gorbachev', for the U.S. 'Smith' or 'Adams'.
The inapprobriate way to match spelling and pronouncing with the Soundex system (which only works with english names) just adds to the problem. In french names you often can't make a difference between 'en', 'an', 'on', 'ent', 'ant', and 'ont'. Famous are the words 'son' (his/her, but also tone), 'sont' ([they] are), 'sans' (without), 'sang' (blood), which are all pronounced the same.
In german names there is no difference between 'tz' and 'z' or between 't', 'tt' and 'dt'. For my family name I know of the spellings 'Sigmund', 'Siegmund', 'Siegmundt', 'Sigemund', 'Siegemund', 'Siegemundt', 'Sigesmund', 'Siegesmund', 'Siegesmundt', 'Zygmunt', and 'Zygmont' (it appears that no one I ever met was able to spell my name right without me helping him, I have several documents with a falsely spelled name...).
Add to this list the number of falsely entered records (people accused of things, later cleared, but not erased from lists compiled while they were under suspicion, typing errors, missunderstandings...). So I am quite amazed that the SFO airport had only about three hundred false positives. On the other hand there were no rightful positives (no one got convicted of terrorism yet who was caught with the CAPPS system), and we don't know about the false negatives (actual terrorists flying without being caught) because they slipped through the controls.
What stops SCO, or anyone for that matter, from copying some code directly from Linux and then saying that it was theirs to begin with. Sure there are some commit logs, etc...
The fact that IBM has the code also... and they have it since about 20years. So they know exactly when which change was done. And they are the defending party in this case.
You were quite short on reading my posting. I was just saying that with this technique the black box a compiled object with a known interface should be is more a darkish grey instead, where you could make out some shadows, the private members. For some algorithms those information gives you some clue how the object works.
Of course you could always analyze the binary code instead and thus figuring the things you are eager to find out. But knowing about the private methods and variables gives you a shortcut. You still have to analyse what happens inside a method, and you still have to find all the places which affect a variable, but at least you get a basic structure.
I know some languages which are object oriented and compile to byte code instead of a binary which enforce the 'private' modifier more strictly and throw a 'security violation' error if you try to access a private member this way. I also know some ways to trick those languages in violating their own rules, but for most of them you have to know the name of the private member beforehand.
Re:By design?
on
Hijacking .NET
·
· Score: 3, Interesting
I completely fail to see what's nasty about it. If it's not documented, you just plain should not rely on it. Isn't hat the way all code works?
The problem is the same like the old one which let you disassemble VisualBasic programs back to the sourcecode about some years ago.
If you put a trade secret inside your code, it is now much easier to access by an outsider because she doesn't have to rely on the binary code alone, but can spy on the structure of the program. This gives her a pretty good clue which algorithm was used and which data structures were used.
Basicly it makes everything you do in.NET a "half open source" program.
Random mutations, which we assume happen all the time, could just as easily start a new strain of human-killing wheat, without the help of Monsato.
There is just one difference: While it is true that mutations and interbrew could generate potentially poisonous plants, they are not the only plants. But genetically altered seeds are mostly used in monoculture, on many fields at the same time. So instead of a small number of plants growing locally and maybe killing a dozen people because of their poison, you have thousands of acres with poisonous plants being transported everywhere and sold in the stores across the whole country.
The problem mit genetically altered food is not the general danger that comes from unexpected effects of the change. It is the large scale at which those changes affect us. Basicly those tests required by the reglements are nothing else than a shortened version of the natural distribution. First a closed biotope, then a small region, then a country... And at every stage you have to check for the dangers occuring.
Monsanto's claim was originally that he arranged (barter or sale) to have a monsanto-licensed farmer give him some of their roundup-ready seed (in violation of contract). Schmeiser claimed that it had appeared on his land, and he had the right to do what he wanted to with his crop. The (lower) courts decided that it didn't matter how the seed had landed on his land.. Monsanto had a patent on the seed, and nobody not licensed by them was allowed to use seeds with those genetics.
Think this a little further. Think of a second company selling genetically altered canola seed to a farmer, and again some of the seed falls over to a neighbour. But this time this farmer isn't using his own seed but Monsanto's. Then you have a farmer with Monsanto seed contamined by another seed. Which decision should the court make now? Handing over the contamined seed to Monsanto (because it violates Monsanto's patents)? Or handing it over to the other company (because it violates their patents)? Or part it half-by-half and giving 50% to each company? Shall both companies now start to sue each other for violating patents?
Leibniz came up with the idea of "differentials" (extremely small differences) and "integrals" (extremely small sums). Newton called his type of calculus "fluxion method" because his ideas were formed around the term of the "fluxion" which served about the same purpose. Newtons terms proved to be more clumsy than those of Leibniz, nevertheless he accused Leibniz of stealing his ideas even though he didn't publish his method until 1770. The Royal Academy later convicted Leibniz of plagiarism, even though Leibniz published his ideas earlier than Newton. Newton argued that his ideas were fundamental for his Principiae mathematicae (which were indeed published before Leibniz). That's probably why Leibniz stopped publishing later. Interesting fact: Newton was president of the Royal Academy when it convicted Leibniz;)
Do these problems really exist? are they bad? Can we have some real evidence instead of just opinion.
Those problems exist indeed. For instance there are two german words, Masse and Maße, the one meaning the mass, the other the general dimensions. Both words are pronounced differently, but spelled the same once converted to uppercase: MASSE.
The german army had this problem already in the days of the tele typer, which was case insensitive in the way that it only accepted lower case letters and no umlauts and no ß. To avoid messing up "masse" (ex Masse) and "masse" (ex Maße) (which could easily happen because the meaning is quite close but not the same, so often the context allowed both interpretations), the solution at this time was to have the german letter 'ß' translated to 'sz' instead of 'ss'. So the both words became 'masse' and 'masze'. It gets messy of course if you have people with differently spelled names, which look the same once converted uppercase. You can't really distinguish anymore between Mr. Grosz and Mr. Groß. It gets worse if you think that names of polish origin which are quite abundant in Germany have the 'sz' pronounced like 'sh', not like the sharp 's'.
So whatever you do, case insensitive always means to loose information. And with verbal information which is not syntactically uniform, you often loose semantics by omitting structural information.
Slashdot Mirror
But aren't there books, written on paper, describing RLE? Maybe the answer is to store some books describing file formats in an inert atmosphere?
The problem is not to decode RLE files. The problem is to have the idea to interpret them as RLE. Look how the fax machine compresses the data in RLE: It has two colors, white and black. It always assumes, the first color is white and then sends a byte with the number of white dots to follow. Then it sends a byte with the now following number of black dots and repeats this again and again. If for some reason the first dot is black, it starts with a 0-byte (No white dots), then the number of black dots to follow. If there are more than 255 dots of one color following each other, then it sends a FF-byte, then a 0-byte, and then the remaining dots in the next byte.
It is for itself a quite easy algorithm, and it is easy to encode and to decode (fax machines should be reliable and simple).
But imagine you get a file with all the bytes in it. How do you tell it's RLE data encoding a picture? Maybe it's MIDI data instead? Or the index of a filesystem?
The rosetta stone contained the message in Ancient Greek (a dead but widely known language at the time of deciphration), Coptian and Hieroglyphic. Even though Coptian was at least known to some specialists and people able to read Ancient Greek were abundant at the time (and still are), it took about 25years to decipher the hieroglyphic texts.
And this was with a language which itself was very easy mapped to the letters (every consonant mapped to a letter, vowels omitted).
The rules which encode a file may be much more complicated. Look just at the most common compression methods (Run Length for instance), how they just add another layer above the already encoded contents. And they remove something very important for deciphration, the redundancy, out of the data. Then the subjects that are stored in files are much more diverse. We have not only language, we have music and graphics, 3D data and cryptographic certificates, configuration files and program binaries.
Just to be able to know what the file is about and thus have an idea how to get started can prove to be more complicated than any deciphration from archaeologic texts.
That's what I call "the wrong way around". You should have it really stored at the same place on every installation. But to the user you should have a localisation layer, that translates "Program Files" to whatever her locale wants to have. So no program has ever to take care where to store its files, and the localisation layer transforms it to whatever the local user wants.
No, it is C:\Programme\.
What? You are NOT using the german version?
I hate it when programs never check for the default install path and just hardcoded assume, it would be C:\Program Files\.
And I am pretty sure, it would be something like C:\Fiches des Programmes\ in the french version.
Don't tell me, Windows got it right from the start. They messed it up like everyone else. And this in the same company.
A paper ballot and a pen is the only form of ballot I trust. And if they don't count the ballots AT THE POLLING PLACE in plain view of the public BEFORE they ship them off to the court house you can't trust the result.
Are you trying to tell me, that in the U.S. the votes don't get counted in the polling place for each voting district in the open public?
Silly me to think it would be a prerequisite for a democratic election to have a local public control how the votes are counted. But on the other hand: We surely have to learn from the U.S. how to screw up a simple thing like an election by trying to automatise everything.
Yes I know you can also screw up while trying to vote with paper and pen. That happens. If it happens to you just write on the paper "INVALID VOTE" after miscasting your vote or if you think your vote could be missinterpreted because you didn't know how to make a cross at the right place. At least you know it's only you who screwed up your vote.
You only have to release source code if you -distribute- your modified kernel.
To be more specific: You have only to release the source code to the people you are distributing the binaries to. So if someone modifies the Linux kernel and gives the binary kernel to someone else, he has to make available the source code of the modified kernel in the same way to exactly the same person.
What you have to permit the recipient of both binary and source code is to do the same, modify the source, create new binaries and give them away to someone else together with the newly modified source code.
Those other ones are not necessarily the common public, they are just the people you are giving your modifications to. So GPL doesn't require you to publish your source. It requires you to make your changes in the source available in the same way you make the modified binaries available. If the target audience is just a small group of people, then the modified source is only available to the same small group of people. If you using the modifications inhouse, the modified source can stay inhouse.
Please order "Wiener Schnitzel mit Spaetzle".
Spindler spots jurisdictional uncertainties for all parties involved: Developers may be held liable if software does not work as expected, even if they only participated marginally in the development, rather than being a lead developer. Employers could walk on thin ice if they pay employees for writing Open Source Software. And buyers of such software must be prepared that liability is limited to the criteria common for items given away for free, i.e. severe negligence only.
But the same is valid for all commercial software in Germany too. EULAs have repeatedly been denied validity because of the german contract law. EULAs are a contract between you and the author of the software. But because you didn't buy the software directly from the author, but from a third party (the reseller, the company bundling the software with a computer etc.pp.), EULAs can't be enforced. All the author can impose on you is the priviledges he gains from the Author's Right (Urheberrecht).
This makes software under GPL in no way different than any commercial software you buy in Germany from a liability point of view.
What you were linking to is the Swisstool, not the Swiss Army Knife. Rather look at this page for reference. Ok, I have to admit that yours look more like a Leatherman Tool after all ;)
That difference is absolutely negligible.
It is not. I did actually read the presentation of the intervention given in the patent statement, and I remember some parts of the presentation we once got.
Here are the main differences to Akamai and Squid:
1. Mirror Image determines which source to use by looking where the request comes from first. This is actually done by having ".mirror." (or another subdomain similar to this one) as part of the requested URL and the subdomain's resolver being part of the Mirror Image system. So the resolver knows where the request comes from and thus triggers the right source to actually deliver the requested information.
2. Squid and Akamai always deliver the content themselves, either from cache or by first requesting the information from the server and delivering it then to the user. Mirror Image instead, if not able to serve the request from cache, sends a 302 Redirect page to the browser (so the browser sends a new request to the original source) and then itself requests the information and stores it in the cache. Thus, Mirror Image claims, the responsivity of the cache increases.
So Squid and Akamai don't have anything to worry about because of this patent. They are operating a different way not covered by the patent.
Having worked with Mirror Image I have to say that the way Mirror Image is doing the caching differs strongly from Akamai's.
While Akamai is putting cache servers in many IP provider's locations (I think more than 5000 so far), Mirror Image is concentrating its caches in about 20 locations connected to the big exchange and peering points. The Mirror Image presenters were explicitely stressing this point and that this other approach is the key to Mirror Image's success. So I guess the patent covers the Mirror Image Way Of Doing Things rather than the idea to cache websites to speed up transfer rates.
Then why don't you submit your results?
You're not disputing the fact that AT&T owned System V at one point in time, are you?
No, he's just disputing AT&T's right to sell (or give away) the termination clause, because this is a mutual agreement and thus only changeable if IBM agrees (as stated clearly in section 4). If AT&T could have assigned the right to any other entity it would have been "AT&T or assignee" in the agreement and not just "AT&T".
It's a lot like the AIDS/HIV test. You want every potential positive test to show up so you can follow it up but you do not want to miss a potentially infected person.
Normally every medical test in fact consists of two tests. The first one is called the efficient test, the second one the effective test.
The first one is to throw out as much uninteresting people (the ones without any sign of sickness) as possible without missing any of the interesting people (the ones that are sick). In Computer Science you call a test like this "Trivial Reject". It should be fast and cheap to save on costs (of every type, like monetary, computational time...).
The second test is supposed to catch as much of the interesting people (sick ones) as possible while throwing out pretty surely all of those who are in fact healthy and thus uninteresting for the matters of the test. This leaves you with a test population which consists of almost all sick people and some very few false positives. Because of the shrunken number of persons this test can be more elaborate and expensive.
The problem with all those tests is, that with the low frequency of the usual dangerous plague you are testing for, the false positives (even though they may be very small compared to the whole of the population) still contains more people than the actual sick ones. I remember some exercises where we had to calculate the risk of actual being sick after a positive test for given frequencies of occurance of the plague in the population and the characteristics of the test. Often the result was that you have a 99% chance of being a false positive even though the test itself was pretty good.
Same is valid for the frequency of the actual terrorists in the population of all U.S. domestic flyers. While there were 19 terrorists in the 9/11 attack, there have been billions of individual flights in the recent years. If your tests are being 99.999% sure in clearing a innocent flyer, it would still mean that this test applied to one billion flight passengers gives you 10 000 false positives.
(This number is one of the reasons why the Romans once decided for "in dubio pro reo". Because the number of lawful citizens is quite high compared to the number of criminals, the average screening of larger populations results in more false positives than actually caught thieves. So you have always to prove individually that someone is guilty beyond any doubt before you can actually call them guilty.)
In this special case the screening for names itself is a very inaccurate test. There are names in some countries being too prevalent to distinguish between people. For Corea it may be the name 'Kim', for India 'Singh', for Russia 'Gorbachev', for the U.S. 'Smith' or 'Adams'.
The inapprobriate way to match spelling and pronouncing with the Soundex system (which only works with english names) just adds to the problem. In french names you often can't make a difference between 'en', 'an', 'on', 'ent', 'ant', and 'ont'. Famous are the words 'son' (his/her, but also tone), 'sont' ([they] are), 'sans' (without), 'sang' (blood), which are all pronounced the same.
In german names there is no difference between 'tz' and 'z' or between 't', 'tt' and 'dt'. For my family name I know of the spellings 'Sigmund', 'Siegmund', 'Siegmundt', 'Sigemund', 'Siegemund', 'Siegemundt', 'Sigesmund', 'Siegesmund', 'Siegesmundt', 'Zygmunt', and 'Zygmont' (it appears that no one I ever met was able to spell my name right without me helping him, I have several documents with a falsely spelled name...).
Add to this list the number of falsely entered records (people accused of things, later cleared, but not erased from lists compiled while they were under suspicion, typing errors, missunderstandings...). So I am quite amazed that the SFO airport had only about three hundred false positives. On the other hand there were no rightful positives (no one got convicted of terrorism yet who was caught with the CAPPS system), and we don't know about the false negatives (actual terrorists flying without being caught) because they slipped through the controls.
And who can mess with TNT? I mean, he's dynamite.
It is less dangerous to mess with TNT (Tri Nitro Toluol) than with dynamite (Tri Nitro (1,2,3)Tri Propanol ester + kieselguhr).
What stops SCO, or anyone for that matter, from copying some code directly from Linux and then saying that it was theirs to begin with. Sure there are some commit logs, etc...
The fact that IBM has the code also... and they have it since about 20years. So they know exactly when which change was done. And they are the defending party in this case.
You were quite short on reading my posting. I was just saying that with this technique the black box a compiled object with a known interface should be is more a darkish grey instead, where you could make out some shadows, the private members. For some algorithms those information gives you some clue how the object works.
Of course you could always analyze the binary code instead and thus figuring the things you are eager to find out. But knowing about the private methods and variables gives you a shortcut. You still have to analyse what happens inside a method, and you still have to find all the places which affect a variable, but at least you get a basic structure.
I know some languages which are object oriented and compile to byte code instead of a binary which enforce the 'private' modifier more strictly and throw a 'security violation' error if you try to access a private member this way. I also know some ways to trick those languages in violating their own rules, but for most of them you have to know the name of the private member beforehand.
I completely fail to see what's nasty about it. If it's not documented, you just plain should not rely on it. Isn't hat the way all code works?
.NET a "half open source" program.
The problem is the same like the old one which let you disassemble VisualBasic programs back to the sourcecode about some years ago.
If you put a trade secret inside your code, it is now much easier to access by an outsider because she doesn't have to rely on the binary code alone, but can spy on the structure of the program. This gives her a pretty good clue which algorithm was used and which data structures were used.
Basicly it makes everything you do in
Random mutations, which we assume happen all the time, could just as easily start a new strain of human-killing wheat, without the help of Monsato.
There is just one difference: While it is true that mutations and interbrew could generate potentially poisonous plants, they are not the only plants. But genetically altered seeds are mostly used in monoculture, on many fields at the same time. So instead of a small number of plants growing locally and maybe killing a dozen people because of their poison, you have thousands of acres with poisonous plants being transported everywhere and sold in the stores across the whole country.
The problem mit genetically altered food is not the general danger that comes from unexpected effects of the change. It is the large scale at which those changes affect us. Basicly those tests required by the reglements are nothing else than a shortened version of the natural distribution. First a closed biotope, then a small region, then a country... And at every stage you have to check for the dangers occuring.
Monsanto's claim was originally that he arranged (barter or sale) to have a monsanto-licensed farmer give him some of their roundup-ready seed (in violation of contract). Schmeiser claimed that it had appeared on his land, and he had the right to do what he wanted to with his crop. The (lower) courts decided that it didn't matter how the seed had landed on his land.. Monsanto had a patent on the seed, and nobody not licensed by them was allowed to use seeds with those genetics.
Think this a little further. Think of a second company selling genetically altered canola seed to a farmer, and again some of the seed falls over to a neighbour. But this time this farmer isn't using his own seed but Monsanto's. Then you have a farmer with Monsanto seed contamined by another seed. Which decision should the court make now? Handing over the contamined seed to Monsanto (because it violates Monsanto's patents)? Or handing it over to the other company (because it violates their patents)? Or part it half-by-half and giving 50% to each company? Shall both companies now start to sue each other for violating patents?
Sorry... the date should have read 1707 ;)
Leibniz came up with the idea of "differentials" (extremely small differences) and "integrals" (extremely small sums). Newton called his type of calculus "fluxion method" because his ideas were formed around the term of the "fluxion" which served about the same purpose. Newtons terms proved to be more clumsy than those of Leibniz, nevertheless he accused Leibniz of stealing his ideas even though he didn't publish his method until 1770. The Royal Academy later convicted Leibniz of plagiarism, even though Leibniz published his ideas earlier than Newton. Newton argued that his ideas were fundamental for his Principiae mathematicae (which were indeed published before Leibniz). That's probably why Leibniz stopped publishing later. Interesting fact: Newton was president of the Royal Academy when it convicted Leibniz ;)
Do these problems really exist? are they bad? Can we have some real evidence instead of just opinion.
Those problems exist indeed. For instance there are two german words, Masse and Maße, the one meaning the mass, the other the general dimensions. Both words are pronounced differently, but spelled the same once converted to uppercase: MASSE.
The german army had this problem already in the days of the tele typer, which was case insensitive in the way that it only accepted lower case letters and no umlauts and no ß. To avoid messing up "masse" (ex Masse) and "masse" (ex Maße) (which could easily happen because the meaning is quite close but not the same, so often the context allowed both interpretations), the solution at this time was to have the german letter 'ß' translated to 'sz' instead of 'ss'. So the both words became 'masse' and 'masze'. It gets messy of course if you have people with differently spelled names, which look the same once converted uppercase. You can't really distinguish anymore between Mr. Grosz and Mr. Groß. It gets worse if you think that names of polish origin which are quite abundant in Germany have the 'sz' pronounced like 'sh', not like the sharp 's'.
So whatever you do, case insensitive always means to loose information. And with verbal information which is not syntactically uniform, you often loose semantics by omitting structural information.
Actually, it was developed by Jim Getty while at MIT.
The same one that is spawning all those getty-processes at my box?
They actually missed their landing point and landed in the Kazakhstan desert
;)
But on the other hand: If they had hit their destination point, they would also have been in the Kazakhian desert. So what's the matter?
(For those interested in a literary approach to the kazakhian landing site, check out The day lasts more than hundred years by Dshingis Aitmatov.)