This is an example of test-driven development. Rather than writing the code and then writing the test cases, you turn it on its head and write the test cases first. You produce the test cases from the requirements -- in this case, the specifications -- and then you're finished development when all of the tests pass.
If you run into a situation where you think the tests should pass and they don't, then there are two possibilities: there's a bug in your code, or there's a bug in the test. You then debug to figure out which is wrong and fix the error. The tests help you find bugs in your code, and your code helps find bugs in the tests.
Large broadcasters such as the BBC in the UK have gone to great lengths to ensure that they can deliver content to users efficiently. They have peering agreements with all of the major ISPs in the UK so that the streaming video (and all of their normal web content, of course) doesn't have to route through anyone else's network. It's interesting to note, though, that their downloadable TV service "iPlayer" does still use P2P. At least in iPlayer's case most people using the content are in the UK and so the peers are relatively close to one another compared to your average warez torrent.
Aside from the obvious reason that I enjoy writing open source software, I also enjoy the fact that I can write the guts of a program and release it and, assuming my program is useful to others, recieve patches from others to make it better. It's also satisfying not to have to reinvent the wheel when I can just make one relatively-minor contribution to an existing project and get a system that meets my requirements.
In an ideal world, they'd create some sort of extensibility framework that allows them to keep their customizations distinct from the "core product", and contribute that to the original project. This way their customizations are not patches but rather plug-ins or customizations that run alongside the core product. As long as that API never changes they can upgrade to any newer version after a suitable amount of integration testing.
Of course, most companies don't wish to invest the time required to do this. I only wish I could convince my employer to do this, so that it wouldn't be a major pain each time we upgrade to another upstream release of a project we depend on.
Related to your email example is that spam filters will generally check to see if a domain transmitting mail actually exists before accepting mail from that domain. This makes it possible for someone sending spam to use any non-existant domain as a source and the spam filters will just see that the domain exists and (presumably) publishes no SPF records and therefore accept the message.
If your requirements are that you want to use patent-encumbered media formats, Microsoft-owned fonts and proprietary network protocols then it's no wonder that you're going to struggle with an operating system that ships only with free software. You might have better luck with something like Mandriva Powerpack.
My father said the same thing to me. I went to a website he uses and used the "Forgot your password?" system. He quickly understood why he should care about people reading his email.
Don't say "in Linux", say "in Ubuntu" or "in Fedora". Ubuntu itself very rarely mentions the name "Linux" in its stock UI. End-users don't need to know that Ubuntu is Linux any more than they need to know that Windows Vista is Windows NT.
I believe that your parent was talking about 10BASE2 -- Ethernet over co-axial cable. While I guess you could in theory do this over twisted pair cables, I'm not sure why you'd want to. Presumably all you'd end up with really is a little gadget at each host that has RJ45 sockets on both sides and an RJ45 plug on the bottom. This little gadget would presumably be functionally the same as a hub.
Telephones have an annoying characteristic in that they have an alarm strapped to them that sounds whenever someone dials your number. Alarms are -- by design -- difficult to ignore and designed to snatch your attention away quickly from whatever other task you are doing. Obviously I can't speak for anyone else, but I find that whenever a phone rings my concentration is broken and so there's little I can do to "finish off" before I answer the phone.
On the other hand, when I get a message in my IM client the notification icon just blinks a bit and I acknowledge it but wait until a convenient moment to actually address it. I've been trying to get my company to get an network-telephony integration system so that I can have a similar notification for my phone, but it would seem that I'm the only one with this problem.
It is correct to say that.NET Reflector is a decompiler. It just happens that IL (the bytecode that.NET languages compile to) is quite high-level, so it's much easier to turn IL into readable C# code than it is to turn x86 machine code into C..NET assemblies also carry a lot of metadata along with the code, allowing Reflector to find the namespaces, classes, properties and methods without needing the original source code. Reflector itself also does some other magic when possible. For example, it can read in the debug information that the C# compiler can optionally produce (the.pdb file) and use that to fill in the things that you can't find out from the assembly alone.
An interesting (for some values of "interesting") exercise is to take some VB.NET code, compile it and then use Reflector to view it in C#. Since the two compilers don't generate completely equivalent code, you can get some unexpected results. A good example to try is some VB.NET code using late binding of methods on variables of type "Object", which isn't a language feature in C#.
If you're looking for an example of a tangible product that has no resale value because it's useless to others, consider address labels. These have your name and address printed on them. Your address labels are useful to you -- you can stick them on the back of an envelope to avoid writing out a return address -- but they are useless to anyone else, because they have a different name and/or address to you.
Other people can of course by their own address labels from the supplier which have their name and address on them, but you won't have much luck selling the printed address labels you already purchased.
Would you argue that address label manufacturers should be required to only produce labels that can be erased and re-written by the end-user? Your recourse in this case would be to only purchase address labels from manufacturers of rewritable address labels. If no such manufacturer exists, you are free to try to start your own business selling rewritable address labels. Of course, the folks that supply your raw materials (in the DRMed music case, the record companies) might just refuse to sell you their product.
Regardless of your legal entitlement to sell something like, say, a DRMed track purchased on iTunes, the publisher isn't required to sell you a product that has value to anyone other than you. You buy DRMed music with the understanding that it's useless to other people, and therefore you're unlikely to be able to sell it. The law isn't stopping you selling it, but the fact that it has no value to anyone but you prevents you from actually finding a buyer.
A hypothetical solution to this problem would be to reverse the relationship and store your document "in" your script, rather than trying to store the script inside the document. This could either take the form of bundling the document as a resource of the script, or more likely just making the script a generic app that can accept arbitrary external input. (An application!)
Spreadsheet formulas are of course a different matter. The notion of a spreadsheet is pretty dependent on the ability to embed code in it, otherwise it'd just be a really poor database engine.
This reminds me of the way the AmigaOS treated floppy disks. Each disk had a label, and if an app tried to do any I/O operation on a label that was not present, the system would pop up a dialog saying something like "Please insert volume DOCUMENTS in any drive". The I/O operation would block until either the disk was inserted, in which case it'd carry out the operation, or the Cancel button was pressed, in which case it'd fail with an error code.
To bring that forward to modern tech, I guess you'd use the UUID of the partition instead of the label, and you'd have to do some sanity checking to make sure that the filesystem hasn't been tampered with by something else before it was re-inserted.
Google Calendar has a synchronization API. I use it to synchronize my mobile phone calendar with my calendar at Google.
I don't know whether or not someone has written a plugin to make Outlook support this protocol, but there's nothing at Google's end stopping them from doing so.
This argument -- that is, the "I want to keep everything separate!" argument -- always seems to come up when stuff like this is under discussion. The important thing to bear in mind is that these technologies are there to enable you to link profiles and transfer data between sites. There's nothing forcing you to do so. If you continue to maintain a separate profile for each site, then you haven't gained nor lost anything.
I currently have a "work" persona and a "personal" persona, plus a whole bunch of vaugely-anonymous personas. These new technologies like OpenID and so on have been a boon for my "personal" persona, but haven't had any effect whatsoever on the one-off "anonymous" personas I have on certain sites, even where those sites have support for these linking/sharing technologies.
Although this is probably pulling the analogy to tearing point, you could probably think of the ".class" files that javac emits as being roughly equivalent to the ".o" files your C compiler might emit. Java doesn't do static linking, so there is no compile-time linker step. The equivalent of your linker is the class loader in the Java runtime.
For those who like the semantics-first approach of LaTeX but don't want to type out formatting codes manually, LyX serves as a good compromise. (some weird bugs notwithstanding.)
Moonlight will be available as a single download from Novell which will include the browser plugin plus the Silverlight graphics engine and the Mono runtime. The codecs necessary to host Silverlight content on Linux will be available from Microsoft. The Moonlight installer will make obtaining the MSFT codecs a seamless step in the process.
My interpretation of this is that Microsoft will be supplying the codecs, but the "support" will take the form of helping Novell to integrate them. Novell will be providing the end-user support resources since the codecs will be distributed as part of Moonlight itself. The codecs are going to be an additional binary blob distributed alongside Novell's release of the otherwise open source Moonlight.
A further point, which sadly I cannot find a reference for right now, is that the codecs were to be licenced only for use as part of Moonlight. This suggests that they are unlikely to be distributed as, say, gstreamer codecs. More likely is that they will have some proprietary API that Mono can call. I would assume, though, that it won't be long before someone writes a legally-shady wrapper that exposes a GStreamer (or other) interface for the codecs in much the same way as currently several DirectShow codecs from Windows are used in Linux builds of Xine right now.
So, wmv is now multiplatform? Are they going to release and support the codecs needed to give full access to the funcionality available on windows-silverlight?
I took my Dell Ubuntu laptop back to my family home this Christmas. I let my younger brother (16 years old) use it a bit to reduce contention on the family PC, which runs Windows XP Home.
My brother quickly got the hang of it. He found Firefox. He found OpenOffice. He made the UI theme for his user account look really ugly and teenager-y, just like on the Windows machine. I realised on subsequent days why this is: kids these days are exposed to all sorts of computer-like tech that doesn't run Windows: mobile phones, handheld consoles, non-handheld consoles, set-top boxes, fancy televisions... they're used to quickly adapting to new user interfaces.
It wouldn't surprise me if ten years from now the precise UI is largely irrelevant, and software and devices will be judged on their functionality and design.
I'm not sure from your comment whether you already know about this, but Google Calendar does have a read/write API that uses its general data API scheme. The Google Calendar provider for Mozilla Thunderbird uses this API and allows you to create new appointments in Thunderbird itself.
Jingle support in third-party Jabber clients isn't really Google's problem!
Slashdot Mirror
Only one of T-mobile UK's data plans allows the use of any VoIP-type service. You have to buy the most expensive plan to be allowed to do this.
I have no idea if they actually enforce it, but it's written in the contract I "signed". (aka agreed to online.)
This is an example of test-driven development. Rather than writing the code and then writing the test cases, you turn it on its head and write the test cases first. You produce the test cases from the requirements -- in this case, the specifications -- and then you're finished development when all of the tests pass.
If you run into a situation where you think the tests should pass and they don't, then there are two possibilities: there's a bug in your code, or there's a bug in the test. You then debug to figure out which is wrong and fix the error. The tests help you find bugs in your code, and your code helps find bugs in the tests.
Large broadcasters such as the BBC in the UK have gone to great lengths to ensure that they can deliver content to users efficiently. They have peering agreements with all of the major ISPs in the UK so that the streaming video (and all of their normal web content, of course) doesn't have to route through anyone else's network. It's interesting to note, though, that their downloadable TV service "iPlayer" does still use P2P. At least in iPlayer's case most people using the content are in the UK and so the peers are relatively close to one another compared to your average warez torrent.
Aside from the obvious reason that I enjoy writing open source software, I also enjoy the fact that I can write the guts of a program and release it and, assuming my program is useful to others, recieve patches from others to make it better. It's also satisfying not to have to reinvent the wheel when I can just make one relatively-minor contribution to an existing project and get a system that meets my requirements.
In an ideal world, they'd create some sort of extensibility framework that allows them to keep their customizations distinct from the "core product", and contribute that to the original project. This way their customizations are not patches but rather plug-ins or customizations that run alongside the core product. As long as that API never changes they can upgrade to any newer version after a suitable amount of integration testing.
Of course, most companies don't wish to invest the time required to do this. I only wish I could convince my employer to do this, so that it wouldn't be a major pain each time we upgrade to another upstream release of a project we depend on.
Related to your email example is that spam filters will generally check to see if a domain transmitting mail actually exists before accepting mail from that domain. This makes it possible for someone sending spam to use any non-existant domain as a source and the spam filters will just see that the domain exists and (presumably) publishes no SPF records and therefore accept the message.
SharpDevelop works pretty well as an IDE for .NET development.
If your requirements are that you want to use patent-encumbered media formats, Microsoft-owned fonts and proprietary network protocols then it's no wonder that you're going to struggle with an operating system that ships only with free software. You might have better luck with something like Mandriva Powerpack.
My father said the same thing to me. I went to a website he uses and used the "Forgot your password?" system. He quickly understood why he should care about people reading his email.
Don't say "in Linux", say "in Ubuntu" or "in Fedora". Ubuntu itself very rarely mentions the name "Linux" in its stock UI. End-users don't need to know that Ubuntu is Linux any more than they need to know that Windows Vista is Windows NT.
I believe that your parent was talking about 10BASE2 -- Ethernet over co-axial cable. While I guess you could in theory do this over twisted pair cables, I'm not sure why you'd want to. Presumably all you'd end up with really is a little gadget at each host that has RJ45 sockets on both sides and an RJ45 plug on the bottom. This little gadget would presumably be functionally the same as a hub.
Telephones have an annoying characteristic in that they have an alarm strapped to them that sounds whenever someone dials your number. Alarms are -- by design -- difficult to ignore and designed to snatch your attention away quickly from whatever other task you are doing. Obviously I can't speak for anyone else, but I find that whenever a phone rings my concentration is broken and so there's little I can do to "finish off" before I answer the phone.
On the other hand, when I get a message in my IM client the notification icon just blinks a bit and I acknowledge it but wait until a convenient moment to actually address it. I've been trying to get my company to get an network-telephony integration system so that I can have a similar notification for my phone, but it would seem that I'm the only one with this problem.
It is correct to say that .NET Reflector is a decompiler. It just happens that IL (the bytecode that .NET languages compile to) is quite high-level, so it's much easier to turn IL into readable C# code than it is to turn x86 machine code into C. .NET assemblies also carry a lot of metadata along with the code, allowing Reflector to find the namespaces, classes, properties and methods without needing the original source code. Reflector itself also does some other magic when possible. For example, it can read in the debug information that the C# compiler can optionally produce (the .pdb file) and use that to fill in the things that you can't find out from the assembly alone.
An interesting (for some values of "interesting") exercise is to take some VB.NET code, compile it and then use Reflector to view it in C#. Since the two compilers don't generate completely equivalent code, you can get some unexpected results. A good example to try is some VB.NET code using late binding of methods on variables of type "Object", which isn't a language feature in C#.
If you're looking for an example of a tangible product that has no resale value because it's useless to others, consider address labels. These have your name and address printed on them. Your address labels are useful to you -- you can stick them on the back of an envelope to avoid writing out a return address -- but they are useless to anyone else, because they have a different name and/or address to you.
Other people can of course by their own address labels from the supplier which have their name and address on them, but you won't have much luck selling the printed address labels you already purchased.
Would you argue that address label manufacturers should be required to only produce labels that can be erased and re-written by the end-user? Your recourse in this case would be to only purchase address labels from manufacturers of rewritable address labels. If no such manufacturer exists, you are free to try to start your own business selling rewritable address labels. Of course, the folks that supply your raw materials (in the DRMed music case, the record companies) might just refuse to sell you their product.
Regardless of your legal entitlement to sell something like, say, a DRMed track purchased on iTunes, the publisher isn't required to sell you a product that has value to anyone other than you. You buy DRMed music with the understanding that it's useless to other people, and therefore you're unlikely to be able to sell it. The law isn't stopping you selling it, but the fact that it has no value to anyone but you prevents you from actually finding a buyer.
A hypothetical solution to this problem would be to reverse the relationship and store your document "in" your script, rather than trying to store the script inside the document. This could either take the form of bundling the document as a resource of the script, or more likely just making the script a generic app that can accept arbitrary external input. (An application!)
Spreadsheet formulas are of course a different matter. The notion of a spreadsheet is pretty dependent on the ability to embed code in it, otherwise it'd just be a really poor database engine.
This reminds me of the way the AmigaOS treated floppy disks. Each disk had a label, and if an app tried to do any I/O operation on a label that was not present, the system would pop up a dialog saying something like "Please insert volume DOCUMENTS in any drive". The I/O operation would block until either the disk was inserted, in which case it'd carry out the operation, or the Cancel button was pressed, in which case it'd fail with an error code.
To bring that forward to modern tech, I guess you'd use the UUID of the partition instead of the label, and you'd have to do some sanity checking to make sure that the filesystem hasn't been tampered with by something else before it was re-inserted.
Google Calendar has a synchronization API. I use it to synchronize my mobile phone calendar with my calendar at Google.
I don't know whether or not someone has written a plugin to make Outlook support this protocol, but there's nothing at Google's end stopping them from doing so.
This argument -- that is, the "I want to keep everything separate!" argument -- always seems to come up when stuff like this is under discussion. The important thing to bear in mind is that these technologies are there to enable you to link profiles and transfer data between sites. There's nothing forcing you to do so. If you continue to maintain a separate profile for each site, then you haven't gained nor lost anything.
I currently have a "work" persona and a "personal" persona, plus a whole bunch of vaugely-anonymous personas. These new technologies like OpenID and so on have been a boon for my "personal" persona, but haven't had any effect whatsoever on the one-off "anonymous" personas I have on certain sites, even where those sites have support for these linking/sharing technologies.
Although this is probably pulling the analogy to tearing point, you could probably think of the ".class" files that javac emits as being roughly equivalent to the ".o" files your C compiler might emit. Java doesn't do static linking, so there is no compile-time linker step. The equivalent of your linker is the class loader in the Java runtime.
For those who like the semantics-first approach of LaTeX but don't want to type out formatting codes manually, LyX serves as a good compromise. (some weird bugs notwithstanding.)
The Moonlight FAQ says:
My interpretation of this is that Microsoft will be supplying the codecs, but the "support" will take the form of helping Novell to integrate them. Novell will be providing the end-user support resources since the codecs will be distributed as part of Moonlight itself. The codecs are going to be an additional binary blob distributed alongside Novell's release of the otherwise open source Moonlight.
A further point, which sadly I cannot find a reference for right now, is that the codecs were to be licenced only for use as part of Moonlight. This suggests that they are unlikely to be distributed as, say, gstreamer codecs. More likely is that they will have some proprietary API that Mono can call. I would assume, though, that it won't be long before someone writes a legally-shady wrapper that exposes a GStreamer (or other) interface for the codecs in much the same way as currently several DirectShow codecs from Windows are used in Linux builds of Xine right now.
Yes.
I took my Dell Ubuntu laptop back to my family home this Christmas. I let my younger brother (16 years old) use it a bit to reduce contention on the family PC, which runs Windows XP Home.
My brother quickly got the hang of it. He found Firefox. He found OpenOffice. He made the UI theme for his user account look really ugly and teenager-y, just like on the Windows machine. I realised on subsequent days why this is: kids these days are exposed to all sorts of computer-like tech that doesn't run Windows: mobile phones, handheld consoles, non-handheld consoles, set-top boxes, fancy televisions... they're used to quickly adapting to new user interfaces.
It wouldn't surprise me if ten years from now the precise UI is largely irrelevant, and software and devices will be judged on their functionality and design.
I'm not sure from your comment whether you already know about this, but Google Calendar does have a read/write API that uses its general data API scheme. The Google Calendar provider for Mozilla Thunderbird uses this API and allows you to create new appointments in Thunderbird itself.
Jingle support in third-party Jabber clients isn't really Google's problem!