It's only a matter of time now until Tivo will go away
FUD!!!!!!!!!!!!
1. Tivo showed their HDTV box at CES 2005, and again at 2006. It isn't a technology issue, it is a wait for the Cable Card standards and certifcation (2.0 which allows up and down stream still isn't finished!)
2. Comcast and Tivo are working together. Comcast is paying Tivo to port their software to the Comcast Motorola box. The status as of CES 2006 was that key functionality was working. Tivo will get recurring revenue for every subscriber that chooses Tivo interface and Tivo and Comcast will split advertising revenue generated through Tivo's advertising services.
3. Business is fine. Rogers (CEO) said during last Growth Conference in December that the company could ad 500k subscribers on break even cash flow. There are plenty of other networks besides Comcast/Dish/DirecTV that cannot afford to build and design their own DVR.
DVR is evolving. Tivo SA2's all have broadband through USB. Series 3 (shown at CES 2006 and rumored to be in beta) will have built in broadband. Generic DVRs do not talk to your network. Play your MP3s, show your picture collections, let you view RSS feeds, etc...
Tivo defends their "Time Warp" patent in Texas this March against Dish. That is one of the patents in the Patent Office's Museum.
Tivo has a 1% churn rate, a rare feat in cable and television services.
I know the other frameworks are in Python, but I don't have a problem with tabs. Can anyone make a comparison about why Ruby on Rails is better or worse than these other two rapid web prototyping frameworks? I haven't coded in Python in 4+ years, has Ruby become significantly better or different than Python for web, text, and database processing?
the article is about amazon building up information about you and your preferences. this leads to "silos" that do not interoperate. can this be shared with ebay, or other sites so that you don't have to teach all media commerce sites what types of things you like? no. perhaps a community based repository for preferences makes more sense. watch dick hardt's message from oscon - it just makes sense.
java's microbenchmarks prove where it doesn't have problems.
the java platform's mantra is "create and use objects". when you compare java to languages that allow objects, but are not heavily object oriented in terms of their supporting framework and libraries, then what you are seeing is the expense of the OO paradigm.
the open source movement should be wary of ibm's "the enemy of my friend is my friend" approach. everyone should remember that at one time ibm was just like microsoft, except in hardware... changing interfaces to force out competition and create vendor lock-in, forcing many companies out of the biz. they were the preinnovators of computing monopoly.
java is a bloated pig, and i recognize it is the most sane implementation of a modern OO language we have right now.
but regarding those benchmarks with java and c, they are all heavily biased towards java. how you say? because they end up timing C's memory management which makes calls directly to the OS. On the java side, we only see grabs from the VM's pre-benchmark-timing-OS-allocated heap space. in the benchmarks, we do not see the time java spends cleaning up, but since the c code makes an explicit call to free, we see c's. in the Fourier test, they were using System.arraycopy for java and doing a for loop to move memory in c (yah, instead of memmove.)
now, this is using java to emulate c, if the comparison were object oriented dev using c++/stl vs java, java would be blown to bits.
They would make money on service contracts... who better to provide you with an embedded solution using BeOS than the BeOS team?
the benefit of GPL would mean 1000 more brains would be writing and porting apps for their OS.
it is a win-win situation for both BeOS the company, computer users, and the open source community.
also... this could be the straw that breaks the xbox and microsoft's stranglehold on the gaming platform since BeOS beat MS in opengl implemenation performance, something linux cannot claim.
1. create a virtual os os.
2. define a virtual os executer's interfaces
3. allow virtual os executers to support the chaining pattern (like unix pipes metaphor)
4. let those familiar with a specific os implement the virtual os executer's interfaces
5. let crazy people build crazy filters (between the virtulized os os and the virtualized os executers it is running) like some of the following:
- allowing an os to save its entire state on a power down command and to reload saved state on power up.
- save x number of states every y minutes, in case of failure or hacking, just reload a previous state and fix the problem before you rearriving at it again.
- allow os context switching by yielding after a low level write to network buffers. the filter can then act as a proxy looking at the low level data, reconstructing the tcp/ip packet and perhaps rerouting it back to local host operating system #2 (or #3...). this would effectively allow programmers to do client/server coding and testing on the same box.
- os morphing. a web site that detects an attack can have another os switch as active. imagine the smile on your friendly neighborhood cracker's face when after tripping the port scan alarm, a windows nt box morphs into a free bsd box.
jim
"a love of possession is a sickness of theirs" -- sitting bull
i'm not so sure you are using continuations in your example.
from what i understand, a continuation is a semantic element that allows you to save a place in the future execution of a block of code. the continuation can then be called (with local vars as parameters) at a later point in time.
We all love GvR, but with that same reasoning, the language would not have threads. I'm actually hoping the digital creation's guys ruffle his feathers a bit and get him more proactive in removing some of the barriers to making python a better software engineering tool. sure it is a great glue, but this language has the capability to morph into much more.
python, i way too dynamic to be tamed by a compiler. there are no static types definitions and many python programs make heavy use of the eval functionality.
Best to wait for python3k for such endeavors when python will most likely get optional static types defintions, interfaces, static scoping, and other features that will make python compiler optimization friendly.
have you guys ever thought about doing proof of concept work in blender (www.blender.nl) it is free, but not open. it would probably give you a big boost in productivity. it fully supports the language python and opengl programming (does not support vendor specific extensions). you could use python to bind to os features and have the full strength of a 3D modelling/animation/simulation package at your fingertips. when you get it the way you want it, you just backwards engineer the api's. after all, getting a framework in place that can express functions of an operating system or other high level algorithimic constructs in a meaningful way will be the hard part.
i'll buy the numeric mapping, but the linear is bunk for me. when your eye slides along that line and is moved up by the larger type of the candidate's party, as they continue to slides across, they corrects as the eye is attracted to the bold dark arrow.
for me, four visual operations occured when i casted a pretend vote: 1. a visual scan of the layout. 2. a vertical scan to find my choice. 3. a horizontal trace to the punch/arrow, and one extra glance to see that the punch above (numeric mapping)was mapped to the arrow to the left.
as far as the gestalt grouping principles go, that could go either way. for me i see a pattern in the very first grouping. the punch hole is located on the horizontal centroid.
one addition that would have helped with tracking in their antiquated system would have been to have the borders collapse to the punch.
A few FACTS (given that the news source is correct - link below) that paint a completely different picture.
"Buchanan received 1 percent of the vote in Palm Beach County."
"In all the counties in Florida where there is significant Independent Party membership, Buchanan got a similiar 1 percent."
"Palm Beach County, on the other hand, has a whopping 14,551 members of the Independent Party.
In fact, it has the highest Independent registration in Florida."
and... regarding the statistics used. it is looking at the ratio of bush/buchanan. since there is a relationship between candidates ( a vote for buch is not a vote for bush or gore ) the ratio will show an exaggerated spike in counties where bush does not do well ( because the ratio of bush/buch relationship is being measured ). this is obviously one of the better statistical 'tricks' that could be used to make this datapoint stand out.
the real tragedy... is that the technology is here. each branch of government needs their own slashdot site. Let 'we the people' do the moderating and issue posting. they can then put *our* words into law, listen to *our* needs, but most importantly *we* get role in government that extends beyond picking one of the two electable losers every 4 years.
"""No -- DC is not interested in branding or monetizing Python, so there will be no pressure for marketing gimmicks like that. Python will remain Python, soon to be owned py the Python Software Foundation (more about that idea another time).""" - Guido van Rosssum
this is a very good move in my opinion. the designers and implementors of the langauge will be couped up with one of their biggest customers and users. only good things can come from all the cross pollination. the zope gurus will no doubt influence python's future in the area of backend server capability.
digicool's zope team also teamed up with mozilla (http://www.zope.org/Products/ZopeStudio) to develop a new managment front end that runs using mozilla's new widgets. this experience may come in handy as the python team decides the fate of the default GUI that ships with the core python release, given that the future of tcl is shady (http://www.ajubasolutions.com/company/whatsnew.ht ml) from a support standpoint, as well as that many are unhappy with inelegant solution of going through another tcl to get to tk.
Kudos to GvR for putting the protection of the python and the python community first.
if you code your java like you should, it will be slower than c. think about it, java encourages object granularity and abstraction, these come at a cost. this will also cost you much more memory to store the meta data a jvm keeps at the class and object level.
on to the benchmarks... the two tests a jvm did win were tight looped calculations (one involving recursion which the c compiler does not optimize for). given that x86 processors have seperate cache for data and for code, in a tight loop, a jit will not be re-interpreted and will actually just call the compiled function which in a tight loop is not likely to have even left the cache. hardly, a condition that would be seen in the real world except maybe by a codec and those often get rewritten in assembler to gain access to simd instructions.
for that particular bench, had c++ been invited to the show, a template could have computed the fibonacci series at compile time, thus blowing them all away!
you should also note that intel'c C compiler was not used, and i've seen other benchmarks where it blow away most of the other C compilers by 10-20%.
Yes, I agree that Europe has it correct on this front.
We have no leadership in this area either, just the tiring, suppressive forces of an inefficient, bloated government that is bullied and manipulated by powerful, efficient, self-serving corporations.
Our weakened masses are easily herded by the billion dollar marketing/advertising research agencies. Many, it seems, have chosen the unimpeded right to consume as an outlet, and a way to "turn off and tune out", rather than deal with reality.
The end result is a country that ultimately consumes itself to death, relegating problem solving to the "next" generation.
This is a bit of a rant, but there is a little truth in all ranting.
you're interested in compliance laws and not the real issue of securing the theft of financially sensitive data.
by watermarking the data i don't see how you've gone the extra mile. you're more interested in setting a trap than just moving the data out of reach. besides if you only watermark binaries, any text (uuencode/uudecode) will make it through.
technically, the only place a watermark makes sense, would be in your own internal applications, if the application provided a screen shot capability.
the obvious solution is for your applications department to use an encrypted database, and to restrict the use of applications. any requests for highly sensitive data beyond the insensitive client information should result in a log entry (and the app should let them know it is logging). if you have any applications that run mult-client reports or dump sensitive data, those should be requests that are actually run by your security team.
now the only person you need to watch is the db admin.
my fear is this has nothing to do with RMS and everything to do with the fact the beopen.com writes Guido's check. doesn't anyone else think it wrong that python's new financial backer is calling the shots on issues this critical to python's user community and future success?
GvR, i'm not so sure that the python developer community shouldn't have some say in the matter.
for instance, will python's spread slow if it can't be included in commercial applications? what about the commercial applications that currently shipping or in development that include python because they bought into python's BSD spirited license?
as a python developer and advocate, i would feel much better hearing from Guido (and not beopen.com) why the BSD license has become unpythonic.
Slashdot Mirror
FUD!!!!!!!!!!!!
1. Tivo showed their HDTV box at CES 2005, and again at 2006. It isn't a technology issue, it is a wait for the Cable Card standards and certifcation (2.0 which allows up and down stream still isn't finished!)
2. Comcast and Tivo are working together. Comcast is paying Tivo to port their software to the Comcast Motorola box. The status as of CES 2006 was that key functionality was working. Tivo will get recurring revenue for every subscriber that chooses Tivo interface and Tivo and Comcast will split advertising revenue generated through Tivo's advertising services.
http://news.com.com/TiVo,+Comcast+reach+DVR+deal/
3. Business is fine. Rogers (CEO) said during last Growth Conference in December that the company could ad 500k subscribers on break even cash flow. There are plenty of other networks besides Comcast/Dish/DirecTV that cannot afford to build and design their own DVR.
DVR is evolving. Tivo SA2's all have broadband through USB. Series 3 (shown at CES 2006 and rumored to be in beta) will have built in broadband. Generic DVRs do not talk to your network. Play your MP3s, show your picture collections, let you view RSS feeds, etc...
Tivo defends their "Time Warp" patent in Texas this March against Dish. That is one of the patents in the Patent Office's Museum.
Tivo has a 1% churn rate, a rare feat in cable and television services.
Tivo is going worldwide (TGC).
I know the other frameworks are in Python, but I don't have a problem with tabs. Can anyone make a comparison about why Ruby on Rails is better or worse than these other two rapid web prototyping frameworks? I haven't coded in Python in 4+ years, has Ruby become significantly better or different than Python for web, text, and database processing?
http://turbogears.org/download/
http://www.djangoproject.com/
Thanks to anyone who has used all 3 frameworks or has any insight.
Jim
the article is about amazon building up information about you and your preferences. this leads to "silos" that do not interoperate. can this be shared with ebay, or other sites so that you don't have to teach all media commerce sites what types of things you like? no. perhaps a community based repository for preferences makes more sense. watch dick hardt's message from oscon - it just makes sense.
http://www.identity20.com/media/OSCON2005/
http://www.boingboing.net/2005/10/03/identity_20_t houghtp.html
java's microbenchmarks prove where it doesn't have problems.
the java platform's mantra is "create and use objects". when you compare java to languages that allow objects, but are not heavily object oriented in terms of their supporting framework and libraries, then what you are seeing is the expense of the OO paradigm.
what you are saying doesn't jive with the netbeans configuration file:
netbeans_default_options="-J-Xms32m -J-Xmx128m -J-XX:PermSize=32m -J-XX:MaxPermSize=96m -J-Xverify:none"
which is guarenteed to only give the jvm a maximum of 128 megabytes.
ibm has been contributing to apache for quite some time.
a ch e.html
http://www.linuxworld.com/linuxworld/expo/lw-ap
the open source movement should be wary of ibm's "the enemy of my friend is my friend" approach. everyone should remember that at one time ibm was just like microsoft, except in hardware... changing interfaces to force out competition and create vendor lock-in, forcing many companies out of the biz. they were the preinnovators of computing monopoly.
jim
I'm pretty sure BSD does not.
jim
java is a bloated pig, and i recognize it is the most sane implementation of a modern OO language we have right now.
but regarding those benchmarks with java and c, they are all heavily biased towards java. how you say? because they end up timing C's memory management which makes calls directly to the OS. On the java side, we only see grabs from the VM's pre-benchmark-timing-OS-allocated heap space. in the benchmarks, we do not see the time java spends cleaning up, but since the c code makes an explicit call to free, we see c's. in the Fourier test, they were using System.arraycopy for java and doing a for loop to move memory in c (yah, instead of memmove.)
now, this is using java to emulate c, if the comparison were object oriented dev using c++/stl vs java, java would be blown to bits.
jim
They would make money on service contracts... who better to provide you with an embedded solution using BeOS than the BeOS team?
the benefit of GPL would mean 1000 more brains would be writing and porting apps for their OS.
it is a win-win situation for both BeOS the company, computer users, and the open source community.
also... this could be the straw that breaks the xbox and microsoft's stranglehold on the gaming platform since BeOS beat MS in opengl implemenation performance, something linux cannot claim.
j. herber
1. create a virtual os os.
2. define a virtual os executer's interfaces
3. allow virtual os executers to support the chaining pattern (like unix pipes metaphor)
4. let those familiar with a specific os implement the virtual os executer's interfaces
5. let crazy people build crazy filters (between the virtulized os os and the virtualized os executers it is running) like some of the following:
- allowing an os to save its entire state on a power down command and to reload saved state on power up.
- save x number of states every y minutes, in case of failure or hacking, just reload a previous state and fix the problem before you rearriving at it again.
- allow os context switching by yielding after a low level write to network buffers. the filter can then act as a proxy looking at the low level data, reconstructing the tcp/ip packet and perhaps rerouting it back to local host operating system #2 (or #3...). this would effectively allow programmers to do client/server coding and testing on the same box.
- os morphing. a web site that detects an attack can have another os switch as active. imagine the smile on your friendly neighborhood cracker's face when after tripping the port scan alarm, a windows nt box morphs into a free bsd box.
jim
"a love of possession is a sickness of theirs" -- sitting bull
i'm not so sure you are using continuations in your example.
e r.htm
from what i understand, a continuation is a semantic element that allows you to save a place in the future execution of a block of code. the continuation can then be called (with local vars as parameters) at a later point in time.
http://www.tismer.com/research/stackless/spcpap
jim
We all love GvR, but with that same reasoning, the language would not have threads. I'm actually hoping the digital creation's guys ruffle his feathers a bit and get him more proactive in removing some of the barriers to making python a better software engineering tool. sure it is a great glue, but this language has the capability to morph into much more.
jim
python, i way too dynamic to be tamed by a compiler. there are no static types definitions and many python programs make heavy use of the eval functionality.
Best to wait for python3k for such endeavors when python will most likely get optional static types defintions, interfaces, static scoping, and other features that will make python compiler optimization friendly.
jim
have you guys ever thought about doing proof of concept work in blender (www.blender.nl) it is free, but not open. it would probably give you a big boost in productivity. it fully supports the language python and opengl programming (does not support vendor specific extensions). you could use python to bind to os features and have the full strength of a 3D modelling/animation/simulation package at your fingertips. when you get it the way you want it, you just backwards engineer the api's. after all, getting a framework in place that can express functions of an operating system or other high level algorithimic constructs in a meaningful way will be the hard part.
blender also has a community of thousands.
jim
i'll buy the numeric mapping, but the linear is bunk for me. when your eye slides along that line and is moved up by the larger type of the candidate's party, as they continue to slides across, they corrects as the eye is attracted to the bold dark arrow.
for me, four visual operations occured when i casted a pretend vote: 1. a visual scan of the layout. 2. a vertical scan to find my choice. 3. a horizontal trace to the punch/arrow, and one extra glance to see that the punch above (numeric mapping)was mapped to the arrow to the left.
as far as the gestalt grouping principles go, that could go either way. for me i see a pattern in the very first grouping. the punch hole is located on the horizontal centroid.
one addition that would have helped with tracking in their antiquated system would have been to have the borders collapse to the punch.
jim
A few FACTS (given that the news source is correct - link below) that paint a completely different picture.
1 /8/161334.shtml
"Buchanan received 1 percent of the vote in Palm Beach County."
"In all the counties in Florida where there is significant Independent Party membership, Buchanan got a similiar 1 percent."
"Palm Beach County, on the other hand, has a whopping 14,551 members of the Independent Party.
In fact, it has the highest Independent registration in Florida."
http://www.newsmax.com/archives/articles/2000/1
and... regarding the statistics used. it is looking at the ratio of bush/buchanan. since there is a relationship between candidates ( a vote for buch is not a vote for bush or gore ) the ratio will show an exaggerated spike in counties where bush does not do well ( because the ratio of bush/buch relationship is being measured ). this is obviously one of the better statistical 'tricks' that could be used to make this datapoint stand out.
the real tragedy... is that the technology is here. each branch of government needs their own slashdot site. Let 'we the people' do the moderating and issue posting. they can then put *our* words into law, listen to *our* needs, but most importantly *we* get role in government that extends beyond picking one of the two electable losers every 4 years.
j. herber
"""No -- DC is not interested in branding or monetizing Python, so there will be no pressure for marketing gimmicks like that. Python will remain Python, soon to be owned py the Python Software Foundation (more about that idea another time).""" - Guido van Rosssum
2 .1&mhitnum=10&CONTEXT=972744081.2060845082
t ml) from a support standpoint, as well as that many are unhappy with inelegant solution of going through another tcl to get to tk.
http://x71.deja.com/threadmsg_ct.xp?AN=68684064
this is a very good move in my opinion. the designers and implementors of the langauge will be couped up with one of their biggest customers and users. only good things can come from all the cross pollination. the zope gurus will no doubt influence python's future in the area of backend server capability.
digicool's zope team also teamed up with mozilla (http://www.zope.org/Products/ZopeStudio) to develop a new managment front end that runs using mozilla's new widgets. this experience may come in handy as the python team decides the fate of the default GUI that ships with the core python release, given that the future of tcl is shady (http://www.ajubasolutions.com/company/whatsnew.h
Kudos to GvR for putting the protection of the python and the python community first.
jherber
2 more projects trying to solve that problem:
. uci.edu/juice/intro.html+juice+and+pascal& hl=en
http://www.cminusminus.org/
http://www.google.com/search?q=cache:caesar.ics
jim
java is cool, but your logic is off.
if you code your java like you should, it will be slower than c. think about it, java encourages object granularity and abstraction, these come at a cost. this will also cost you much more memory to store the meta data a jvm keeps at the class and object level.
on to the benchmarks... the two tests a jvm did win were tight looped calculations (one involving recursion which the c compiler does not optimize for). given that x86 processors have seperate cache for data and for code, in a tight loop, a jit will not be re-interpreted and will actually just call the compiled function which in a tight loop is not likely to have even left the cache. hardly, a condition that would be seen in the real world except maybe by a codec and those often get rewritten in assembler to gain access to simd instructions.
for that particular bench, had c++ been invited to the show, a template could have computed the fibonacci series at compile time, thus blowing them all away!
you should also note that intel'c C compiler was not used, and i've seen other benchmarks where it blow away most of the other C compilers by 10-20%.
jim
http://www.google.com/search?q=cache:www.microproc essor.sscc.ru/Merced/+merced+intel+1996& hl=en
HP and INTEL began joint intial R&D for 64 bit computing in 1994.
in 1997 INTEL announced they had the merced (itanium) running via software simulation.
Do you think we will see IA-64 itaniums this year?
LOL, maybe transmeta was being generous.
jim
Yes, I agree that Europe has it correct on this front.
We have no leadership in this area either, just the tiring, suppressive forces of an inefficient, bloated government that is bullied and manipulated by powerful, efficient, self-serving corporations.
Our weakened masses are easily herded by the billion dollar marketing/advertising research agencies. Many, it seems, have chosen the unimpeded right to consume as an outlet, and a way to "turn off and tune out", rather than deal with reality.
The end result is a country that ultimately consumes itself to death, relegating problem solving to the "next" generation.
This is a bit of a rant, but there is a little truth in all ranting.
jim
you're interested in compliance laws and not the real issue of securing the theft of financially sensitive data.
by watermarking the data i don't see how you've gone the extra mile. you're more interested in setting a trap than just moving the data out of reach. besides if you only watermark binaries, any text (uuencode/uudecode) will make it through.
technically, the only place a watermark makes sense, would be in your own internal applications, if the application provided a screen shot capability.
the obvious solution is for your applications department to use an encrypted database, and to restrict the use of applications. any requests for highly sensitive data beyond the insensitive client information should result in a log entry (and the app should let them know it is logging). if you have any applications that run mult-client reports or dump sensitive data, those should be requests that are actually run by your security team.
now the only person you need to watch is the db admin.
jim
the linux today GvR response clears up my questions.
apologies beopen.com folks.
jim
my fear is this has nothing to do with RMS and everything to do with the fact the beopen.com writes Guido's check. doesn't anyone else think it wrong that python's new financial backer is calling the shots on issues this critical to python's user community and future success?
GvR, i'm not so sure that the python developer community shouldn't have some say in the matter.
for instance, will python's spread slow if it can't be included in commercial applications? what about the commercial applications that currently shipping or in development that include python because they bought into python's BSD spirited license?
as a python developer and advocate, i would feel much better hearing from Guido (and not beopen.com) why the BSD license has become unpythonic.
jim