Interesting. If it is true that everything is in raw ATM at Free.fr, I suppose that they wrote some kind of a communication layer for making requests to the DSLAM. If you know of a place where to find further information, I'm interested.
I suppose that you mean multicast download, otherwise it would have no significant benefit over unicast streaming. That videos are "pushed" as files to set-top boxes certainly makes sense. I'm sure there's a deployed system somewhere that makes use of that.
However, for really big numbers of subscribers, it seems to me that having the price of the STBs as low as $80-100 is important. Having a HDD on the box makes it difficult to reach the targeted price range.
Also, if using multicast download, consider these two use-cases:
(1) A good amount of movies that users might want to see are pushed to STBs. Then, there is globally a waste of bandwidth. What if I had my STB switched-off while movies were pushed? How long will it take before I get the new movies?
(2) Movies are multicast-downloaded only when 1 or more users request them. Then, it will introduce a delay. For example, you order a movie you want to watch right away but it will be available in three hours.
I have difficulties thinking about movie download in a way that makes it best for the network, for the IPTV vendor, and for the subscriber.
I think that if IPTV is deployed correctly, the overall cost should be way lower than any solution involved TiVo-like devices.
The idea of watching whatever you want, whenever you want is definitely exciting. As part of my job, I have been studying the possibility to watch any program one would have missed in the past week of programming. You don't even bother to record the programs! All are recorded, on the server side!
However, a big, big factor to consider is scalability. I don't think you can let 100% of IPTV subscribers watch TV at their own pace. The network cannot handle that.
What happens when designing an IPTV system, is that you make guesses about what percentage of users will want to watch live TV and what percentage will watch on-demand, or will want to control live, at a same time.
Live TV works in multicast, and NVOD (Network Video on Demand) works in unicast. If 10 people watch the same show -- but want fine control over it -- that's 10 unicast streams, instead of of 10 "joins" to a single multicast stream. In other words, if you give everyone control, you multiply the bandwidth that is required by ten (in this particular case).
I'm sure that in the future IPTV providers go towards more unicast, on-demand connections; but there will always be live and multicast because that's the best use of the bandwidth.
> First, why do you need IP for TV ? over ADSL, it's a lot better to send it in raw ATM
What everyone expects from TV over DSL is interactivity. People want video-on-demand, they want to pause or record live, they want to provision their account, etc. At the same time, broadcasters and telcos are looking for new revenue streams. Without IPTV, the features mentioned above would only work if you add a hard-drive to each set-top box (STB). However, for deployments to hundreds of thousands of subscribers, the cost per STB is too high.
The two-way communication between the set-top box and the server is what IPTV brings. And for that, HTTP and RTSP are example of protocols to use. (HTTP for sophisticated client/server transactions, and RTSP for fine control over the real-time streaming). I don't know how the sophisticated features would work without the IP, HTTP, RTSP, and RTP layers.
Since apparently you've been investigating Microsoft, I am curious to understand what they truly have.
I watched Microsoft's keynote at CES 2005 and I have been present at several broadcast tradeshows. (Click the link "100 K" or "300 K", the IPTV demo is 47'30 into the stream). I really felt uncomfortable with all the lies. One of them is that by design "IPTV allows instantaneous channel change". What Microsoft is showing is probably a set-top box that decodes 4 streams at the same time. What amount bandwidth does that use? Probably more than you're willing to spend. And, the fact that it's IP doesn't bring anything new here.
What you will never hear from Microsoft is how much power is required to run their applications. I suppose a quite powerful PC.
I'm interested in knowing how does that affect your thinking. Apart from the codec and DRM matters, do you think that Microsoft is a serious actor in the IPTV world?
When you work at Microsoft, the job comes with salary, stock, and among other benefits the possibility for engineers to file patents. It must be very self-satisfying for an engineer to have a patent with his or her name on it.
Maybe it's worth the cost for Microsoft. It would be a way to tell engineers that the company thinks that what they are doing is important. The message only costs money and a lawer's time; it doesn't require much social sophistication.
Note -- This is only a theory. I have never worked at Microsoft, and those I know who did work at Microsoft never told me about this particular fact.
I live in France, and strangely search results start with links to pages in French.
Why would that be wrong? Because of the following:
- I did check that the language setting is to "Search pages written in any language".
- My search request didn't contain any French word.
- My OS is a completely U.S. system. (No French locale)
My conclusion: MSN Search is making too strong an assumption about the fact that I am located in France, but unfortunately I cannot change that.
Since English is the most wide-spread language on the Internet, I am concerned that the French links on top of the results page would not have the best relevance.
Note -- I tried to connect to search.msn.com, but it automatically goes to search.msn.fr. Anyhow, that shouldn't make any difference, as for Google.
If you jump at 26'30 into the video, you will see Gates failing making a demo of Windows Media Center.
At 27'13, he shows no concern whatsoever of what is happening. As Conan O'Brien makes a joke about the situation, Gates is showing total weakness.
At 27'25, Gates finally understands it is his turn to speak. He makes a remark that shows a huge lack of social sophistication: "How to you like this camera?" (Pointing to a Nikon D70 on the table)
There is no need for more evidence than watching any of Gates's public appearances to understand what kind of person he is.
I don't see what we could learn more from a handwriting analysis. There already is enough information available that is more relevant than what is mentioned in the article.
One of the main advantages of MPEG-4 is that it is good at low bit rates and high bit rates.
This is especially true of MPEG-4 part 10, aka H.264. H.264 targets mobile devices and HD-resolution devices. A broadcaster of Japan named MBCo already broadcasts in MPEG-4/H.264 to handheld devices. They are using encoders from Envivio
MPEG-4 part 2 / ASP is older, but also good. 320x240 is definitely large enough a resolution for the encoder. ASP (Advanced Simple Profile) is supported by QuickTime since maybe version 5 (not sure).
As mentioned elsewhere in this thread, it is already possible to use zip compression at transportation.
But there are reasons why XML-specific encoding has chances to be far more efficient. Consider this:
<hello></hello>
For anyone familiar with XML, it translates into:
<hello/>
..which takes less space.
The '<', '>', and '/' represent the "empty element" aspect of the XML code, and that seems like an overkill. Think of way to represent the notion of "empty element". I'm sure that if all notions of XML were listed, you wouldn't need a lot of bits to uniquely code each of them.
Already, without any statistical compression, we've saved many bytes in my example.
Other advantages of being language-specific is that, knowing the weaknesses of the language, the binary format can make a smart use of redondancy. (Such as: I'd rather lose comments than useful code -- may the comments be coded in the binary-XML)
I agree! Also, UI code is not worth dis-assembling
on
Decompiling Java
·
· Score: 1
I agree.
Also, code for user interfaces (UI) is not worth dis-assembling, because most of the intellectual property is visible on screen anyway.
Good UI is the easiest thing to steal in a piece of software.
I don't mean that making better than these programs is easy. However, I think it should be trivial for any user to type a letter or a report.
Discussing the toolbar issue is irrelevant. (Just as a quick note, I think it makes sense to relocate some buttons, given that the format of common documents (Letter, A4) requires more vertical space than horizontal space.)
So the toolbar was just an example. Consider how hard it is to deal with automatic chaptering and numbering in Word. There are many, many other examples.
An advantage of old programs is that they were more predictable, at least.
I don't think a good MS Word replacement should be an MsWord clone. I think OpenOffice is already too much of a Word clone.
It seems to me that AbiWord already suffers from the stuff everything in the toolbar syndrome. Why make the toolbar bar customizable and movable? Because that's what developers think is a fix for the poor usability of their software. In Word, it is easier to trash your toolbar than to type a letter.
Will AbiWord implement Personalized menus, the most stupid workaround to the problem of having too many menu items?
No innovation has been made in the recent years regarding page-layout software usability. That's depressing. I used to teach people how to use Word. Sometimes, I felt like teaching how to troubleshoot.
Programs like Word made the user model too complex and illogical. It has become close to impossible for a user to understand and anticipate how Word functions. I don't think that solely code design and bugs are responsible for that. I think that the user model needs to be completely reviewed.
The article Anatomy of a Software bug quoted in another Slashdot story tells the story of an MsWord bug. A "Write Error" can occur when saving a document after several hours of work. The article says you can workaround the problem by doing your editing in Normal view. (As opposed to Page view(?)--I'm not an MsWord user either.) That may be the problem you noticed.
The bug was created with the introduction of multiple undo/redo in an old design. I understand from the article that large parts of MsWord need to be rewritten. Will that ever happen or will it die? How can software survive more than 10 years at a leader position? Probably not without recurrent and effective code refactoring.
> Which is not exactly true. It could be true but not provable. It could be false but not provable.
"true but not provable" is still makes the equation true. "false but not provable" still makes it false.
That's the logic. Either true or false, not both; regardless of what humans can prove.
On the other hand, the four possible states in our understanding of the equation can be either:
To me, the biggest flaw in AIM is its user interface. It's ugly, it's hard to learn, it's painful to use. I'm sure there's a hundred obvious usability mistakes.
And, why does a company like AOL feels the need to violate my window real estate with ads? (Animated ads!! Movies!!) (Tip to block ads: Set a firewall rule to block any communication with the server ads.web.aol.com)
What is sad is that Gaim doesn't seem to do much better than AIM. Though more efforts were made on the look, the GUI is still messy. (See the menus, the preference dialog, too many dialogs, etc.)
I'm glad you mentioned that. TeX is a programming language. I spent 30 minutes looking for front-ends for TeX, but I found nothing satisfying. Some projects don't even bother showing more than two screenshots.
Let's not forget about one of the most important features for a FrameMaker successor: the graphical user interface!
I think that creating an easy-to-use and yet powerful GUI for TeX can be as challenging as creating TeX itself. I'm convinced that TeX has all the features for text composition that I need, but most technical writers will need a nice GUI.
It looks to me that Adobe has long ago decided to kill FrameMaker.
FrameMaker is very old. It has a Windows 3.11 feeling, and that affects a lot productivity. Dialog windows have an anti-conventional layout. Using undo/redo is often hazardous. There aren't enough keyboard shortcuts. Etc. Also, Adobe has released new versions of FrameMaker without fixing obvious GUI bugs and limitations.
Adobe decided to replace PageMaker with InDesign when PageMaker wasn't that old. But PageMaker had competition: Quark XPress. Without InDesign, Adobe would have lost credibility in the pre-press market.
FrameMaker has powerful features that we need to see in other products. It has sophisticated management table of contents, index, cross references. You can use variables and conditional text. Many comments in this discussion omitted that.
Don't even consider Ms Word and Ms Word clones. They are not optimized for productivity. They lack the features mentioned above. Ms Word is not reliable. Cheap word processors have very poor text justification quality. (Often you can see when a page has been printed with MsWord.)
I don't see InDesign integrating FrameMaker's features. InDesign is not a word processor. In InDesign, you have to explicitly link text from one page to another. Even though there is an automated way of performing this task, it is extra-work when you layout a 200-page manual.
It worries me that Adobe doesn't seem to have plans to replace FrameMaker.
Does anyone have any insight into Adobe's plans?
Side note: I think the only reason PageMaker is still alive is that it remains an easy way for Adobe to earn money. If only 5000 copies are sold every year, that's still $2,500,000 for Adobe. It makes it worth paying a few engineers to add new, but superficial features. Maybe the same thing is happening to FrameMaker.
There is a well-known limitation to QuickTime that makes it impossible to open files which names are longer than 64 characters. I think the defect has been there for a long time.
But, notice something: The source code for that link says my name is "span class='hit' ".
[...] The HTML source code of this result says "span class='d' ".
[...] So, apparently someone at Microsoft saw the article and classified it.
Here are two relevant lines from the CSS file that the search result pages use:
You made a search on "Michael Jennings". As a result, every occurrence of the words "Michael Jennings" are emphasized with bold using the class "hit". The link to the Spanish version of your article doesn't doesn't show your name. That explains why "hit" is not used there.
The class "d" is used to display the text at 12 pixels, which makes it impossible to use my browser's text size settings. (Here is one of Microsoft's abuses.)
Looking at the HTML and the CSS, it seems that class names are not clear because Microsoft tries to save space by making them short. (Also, the HTML has no spacing and no carriage returns.)
Slashdot Mirror
Thanks for the link. I will read the paper since I can read French.
Interesting.
If it is true that everything is in raw ATM at Free.fr, I suppose that they wrote some kind of a communication layer for making requests to the DSLAM. If you know of a place where to find further information, I'm interested.
> why not have the show download to the TV box?
I suppose that you mean multicast download, otherwise it would have no significant benefit over unicast streaming. That videos are "pushed" as files to set-top boxes certainly makes sense. I'm sure there's a deployed system somewhere that makes use of that.
However, for really big numbers of subscribers, it seems to me that having the price of the STBs as low as $80-100 is important. Having a HDD on the box makes it difficult to reach the targeted price range.
Also, if using multicast download, consider these two use-cases:
(1) A good amount of movies that users might want to see are pushed to STBs. Then, there is globally a waste of bandwidth. What if I had my STB switched-off while movies were pushed? How long will it take before I get the new movies?
(2) Movies are multicast-downloaded only when 1 or more users request them. Then, it will introduce a delay. For example, you order a movie you want to watch right away but it will be available in three hours.
I have difficulties thinking about movie download in a way that makes it best for the network, for the IPTV vendor, and for the subscriber.
I think that if IPTV is deployed correctly, the overall cost should be way lower than any solution involved TiVo-like devices.
The idea of watching whatever you want, whenever you want is definitely exciting. As part of my job, I have been studying the possibility to watch any program one would have missed in the past week of programming. You don't even bother to record the programs! All are recorded, on the server side!
However, a big, big factor to consider is scalability. I don't think you can let 100% of IPTV subscribers watch TV at their own pace. The network cannot handle that.
What happens when designing an IPTV system, is that you make guesses about what percentage of users will want to watch live TV and what percentage will watch on-demand, or will want to control live, at a same time.
Live TV works in multicast, and NVOD (Network Video on Demand) works in unicast. If 10 people watch the same show -- but want fine control over it -- that's 10 unicast streams, instead of of 10 "joins" to a single multicast stream. In other words, if you give everyone control, you multiply the bandwidth that is required by ten (in this particular case).
I'm sure that in the future IPTV providers go towards more unicast, on-demand connections; but there will always be live and multicast because that's the best use of the bandwidth.
> First, why do you need IP for TV ? over ADSL, it's a lot better to send it in raw ATM
What everyone expects from TV over DSL is interactivity. People want video-on-demand, they want to pause or record live, they want to provision their account, etc. At the same time, broadcasters and telcos are looking for new revenue streams. Without IPTV, the features mentioned above would only work if you add a hard-drive to each set-top box (STB). However, for deployments to hundreds of thousands of subscribers, the cost per STB is too high.
The two-way communication between the set-top box and the server is what IPTV brings. And for that, HTTP and RTSP are example of protocols to use. (HTTP for sophisticated client/server transactions, and RTSP for fine control over the real-time streaming). I don't know how the sophisticated features would work without the IP, HTTP, RTSP, and RTP layers.
Since apparently you've been investigating Microsoft, I am curious to understand what they truly have.
I watched Microsoft's keynote at CES 2005 and I have been present at several broadcast tradeshows. (Click the link "100 K" or "300 K", the IPTV demo is 47'30 into the stream).
I really felt uncomfortable with all the lies. One of them is that by design "IPTV allows instantaneous channel change". What Microsoft is showing is probably a set-top box that decodes 4 streams at the same time. What amount bandwidth does that use? Probably more than you're willing to spend. And, the fact that it's IP doesn't bring anything new here.
What you will never hear from Microsoft is how much power is required to run their applications. I suppose a quite powerful PC.
I'm interested in knowing how does that affect your thinking. Apart from the codec and DRM matters, do you think that Microsoft is a serious actor in the IPTV world?
When you work at Microsoft, the job comes with salary, stock, and among other benefits the possibility for engineers to file patents. It must be very self-satisfying for an engineer to have a patent with his or her name on it.
Maybe it's worth the cost for Microsoft. It would be a way to tell engineers that the company thinks that what they are doing is important. The message only costs money and a lawer's time; it doesn't require much social sophistication.
Note -- This is only a theory. I have never worked at Microsoft, and those I know who did work at Microsoft never told me about this particular fact.
I live in France, and strangely search results start with links to pages in French.
Why would that be wrong? Because of the following:
- I did check that the language setting is to "Search pages written in any language".
- My search request didn't contain any French word.
- My OS is a completely U.S. system. (No French locale)
My conclusion: MSN Search is making too strong an assumption about the fact that I am located in France, but unfortunately I cannot change that.
Since English is the most wide-spread language on the Internet, I am concerned that the French links on top of the results page would not have the best relevance.
Note -- I tried to connect to search.msn.com, but it automatically goes to search.msn.fr. Anyhow, that shouldn't make any difference, as for Google.
There is already enough evidence that Bill Gates doesn't have the qualities of a leader, and that he has few social abilities.
For example, look at Microsoft's keynote at CES. (Click on 100K or 300K to watch the video.)
There is no need for more evidence than watching any of Gates's public appearances to understand what kind of person he is.
I don't see what we could learn more from a handwriting analysis. There already is enough information available that is more relevant than what is mentioned in the article.
One of the main advantages of MPEG-4 is that it is good at low bit rates and high bit rates.
This is especially true of MPEG-4 part 10, aka H.264. H.264 targets mobile devices and HD-resolution devices. A broadcaster of Japan named MBCo already broadcasts in MPEG-4/H.264 to handheld devices. They are using encoders from Envivio
MPEG-4 part 2 / ASP is older, but also good. 320x240 is definitely large enough a resolution for the encoder. ASP (Advanced Simple Profile) is supported by QuickTime since maybe version 5 (not sure).
As mentioned elsewhere in this thread, it is already possible to use zip compression at transportation.
But there are reasons why XML-specific encoding has chances to be far more efficient. Consider this:
<hello></hello>
For anyone familiar with XML, it translates into:
<hello/>
The '<', '>', and '/' represent the "empty element" aspect of the XML code, and that seems like an overkill. Think of way to represent the notion of "empty element". I'm sure that if all notions of XML were listed, you wouldn't need a lot of bits to uniquely code each of them.
Already, without any statistical compression, we've saved many bytes in my example.
Other advantages of being language-specific is that, knowing the weaknesses of the language, the binary format can make a smart use of redondancy. (Such as: I'd rather lose comments than useful code -- may the comments be coded in the binary-XML)
I agree.
Also, code for user interfaces (UI) is not worth dis-assembling, because most of the intellectual property is visible on screen anyway.
Good UI is the easiest thing to steal in a piece of software.
I don't mean that making better than these programs is easy. However, I think it should be trivial for any user to type a letter or a report.
Discussing the toolbar issue is irrelevant. (Just as a quick note, I think it makes sense to relocate some buttons, given that the format of common documents (Letter, A4) requires more vertical space than horizontal space.)
So the toolbar was just an example. Consider how hard it is to deal with automatic chaptering and numbering in Word. There are many, many other examples.
An advantage of old programs is that they were more predictable, at least.
> is AbiWord a worthy MS Word replacement?
I don't think a good MS Word replacement should be an MsWord clone. I think OpenOffice is already too much of a Word clone.
It seems to me that AbiWord already suffers from the stuff everything in the toolbar syndrome. Why make the toolbar bar customizable and movable? Because that's what developers think is a fix for the poor usability of their software. In Word, it is easier to trash your toolbar than to type a letter.
Will AbiWord implement Personalized menus, the most stupid workaround to the problem of having too many menu items?
No innovation has been made in the recent years regarding page-layout software usability. That's depressing. I used to teach people how to use Word. Sometimes, I felt like teaching how to troubleshoot.
Programs like Word made the user model too complex and illogical. It has become close to impossible for a user to understand and anticipate how Word functions. I don't think that solely code design and bugs are responsible for that. I think that the user model needs to be completely reviewed.
The article Anatomy of a Software bug quoted in another Slashdot story tells the story of an MsWord bug. A "Write Error" can occur when saving a document after several hours of work. The article says you can workaround the problem by doing your editing in Normal view. (As opposed to Page view(?)--I'm not an MsWord user either.) That may be the problem you noticed.
The bug was created with the introduction of multiple undo/redo in an old design. I understand from the article that large parts of MsWord need to be rewritten. Will that ever happen or will it die? How can software survive more than 10 years at a leader position? Probably not without recurrent and effective code refactoring.
> Which is not exactly true. It could be true but not provable. It could be false but not provable.
"true but not provable" is still makes the equation true. "false but not provable" still makes it false.
That's the logic. Either true or false, not both; regardless of what humans can prove.
On the other hand, the four possible states in our understanding of the equation can be either:
To me, the biggest flaw in AIM is its user interface. It's ugly, it's hard to learn, it's painful to use. I'm sure there's a hundred obvious usability mistakes.
And, why does a company like AOL feels the need to violate my window real estate with ads? (Animated ads!! Movies!!)
(Tip to block ads: Set a firewall rule to block any communication with the server ads.web.aol.com)
What is sad is that Gaim doesn't seem to do much better than AIM. Though more efforts were made on the look, the GUI is still messy. (See the menus, the preference dialog, too many dialogs, etc.)
Thanks for all the details about TeX.
TeX is a Turing-complete programming language.
I'm glad you mentioned that. TeX is a programming language. I spent 30 minutes looking for front-ends for TeX, but I found nothing satisfying. Some projects don't even bother showing more than two screenshots.
Let's not forget about one of the most important features for a FrameMaker successor: the graphical user interface!
I think that creating an easy-to-use and yet powerful GUI for TeX can be as challenging as creating TeX itself. I'm convinced that TeX has all the features for text composition that I need, but most technical writers will need a nice GUI.
It looks to me that Adobe has long ago decided to kill FrameMaker.
FrameMaker is very old. It has a Windows 3.11 feeling, and that affects a lot productivity. Dialog windows have an anti-conventional layout. Using undo/redo is often hazardous. There aren't enough keyboard shortcuts. Etc. Also, Adobe has released new versions of FrameMaker without fixing obvious GUI bugs and limitations.
Adobe decided to replace PageMaker with InDesign when PageMaker wasn't that old. But PageMaker had competition: Quark XPress. Without InDesign, Adobe would have lost credibility in the pre-press market.
FrameMaker has powerful features that we need to see in other products. It has sophisticated management table of contents, index, cross references. You can use variables and conditional text. Many comments in this discussion omitted that.
Don't even consider Ms Word and Ms Word clones. They are not optimized for productivity. They lack the features mentioned above. Ms Word is not reliable. Cheap word processors have very poor text justification quality. (Often you can see when a page has been printed with MsWord.)
I don't see InDesign integrating FrameMaker's features. InDesign is not a word processor. In InDesign, you have to explicitly link text from one page to another. Even though there is an automated way of performing this task, it is extra-work when you layout a 200-page manual.
It worries me that Adobe doesn't seem to have plans to replace FrameMaker.
Does anyone have any insight into Adobe's plans?
Side note: I think the only reason PageMaker is still alive is that it remains an easy way for Adobe to earn money. If only 5000 copies are sold every year, that's still $2,500,000 for Adobe. It makes it worth paying a few engineers to add new, but superficial features. Maybe the same thing is happening to FrameMaker.
There is a well-known limitation to QuickTime that makes it impossible to open files which names are longer than 64 characters.
I think the defect has been there for a long time.
iTunes uses QuickTime for handling audio files.
But, notice something: The source code for that link says my name is "span class='hit' ".
[...] The HTML source code of this result says "span class='d' ".
[...] So, apparently someone at Microsoft saw the article and classified it.
Here are two relevant lines from the CSS file that the search result pages use:
The full CSS file is here.
You made a search on "Michael Jennings". As a result, every occurrence of the words "Michael Jennings" are emphasized with bold using the class "hit". The link to the Spanish version of your article doesn't doesn't show your name. That explains why "hit" is not used there.
The class "d" is used to display the text at 12 pixels, which makes it impossible to use my browser's text size settings. (Here is one of Microsoft's abuses.)
Looking at the HTML and the CSS, it seems that class names are not clear because Microsoft tries to save space by making them short. (Also, the HTML has no spacing and no carriage returns.)