If wcbrown thought that most Java books are tough on the beginner, he'll be in even worse straits in the J2EE world.
The Java language itself is fairly sparse. The collection of libraries that make up the JRE (Java Runtime Environment) is immense. And java is pretty useless until you learn your way around the APIs.
I like the language in general. The exception handling model is good. The standard APIs are rich. The namespace issues are handled well.
Developing in Java is a pain in the ass, however, due to classpath and packaging issues. The various IDEs help somewhat, but they can also obscure these issues and prevent you from learning them. In addition, everyone knows that vi is the one true editor - and vim provides good syntax highlighting too.
Best advice to a budding java developer is learn to use apache Ant right from the start. Using JUnit early on is also a good idea. Doing small projects with Ant and Java is far easier and more informative than doing projects in java alone or in an IDE.
I helped do a job installing fiber in a Manhattan office tower almost fifteen years ago. It was being used to interconnect the datacom closets on each floor with a central datacom room. I haven't had reason to use it since.
Is it still has tedious to put the connectors on the ends?
When I was doing it, IIRC, the process ran something like this:
Strip the sheath form the fiber.
Epoxy the fiber into a connector, with plenty sticking out from the "business end".
Use a special knife to score the fiber flush with the connector.
Break off the excess fiber.
Attach the connector to a flat round disk which would hold it perfectly perpendicular to a flat surface.
Using increasingly fine grits of "sand paper", polish the end if the fiber perfectly smooth and flush to the connector by rubbing it (and the disk) in a figure-8 pattern.
Kudos to the Contiki folks. I don't think this is useful, but if they enjoyed themselves doing it, all the better for them!
This should give those of you who have an old x86 PC that is too small to run even the smallest of Linux variants, a chance to browse the web, set up a web server, and doing other essential stuff.
If you have an old pre-386 machine around, why not run Minix? That should make a far more useful machine.
As far as I know, the lawters usually get paid regardless of the outcome.
Actually, that is far from true. The details vary in different jurisdictions, but in general personal injury, medical malpractice, products liability, and class action suits like the one we're discussing are done on contigency. The lawyers must pay all expenses out of their own pockets, in addition to their labor. They can only receive payment if they win the case and collect the judgement.
I'm not sure how consistent this is in other jurisdictions and areas of law, but in NY State a personal injury or medical malpractice case by law pays one third of the judgement. It is illegal for an attorney to charge any other fee. In most of these cases, that 33% actually represents less than the normal hourly rate that an attorney would charge for other kinds of work.
My wife used to practice in this area when she worked for another law firm. When she went out on her own, she could not afford to risk the $10,000 - $100,000 from our own pocket to litigate a single case, a case that may take years to be completed.
I've got some ideas about TCO of windows vs. Linux, and I'm going to present them as facts.
In a corporate server environment, Linux can provide applications servers, database hosting, Web Server, and File & Print services through a suite of truly state of the art packages (JBoss, MySQL, Apache, Samba) for the cost of hardware and a few highly skilled admin people.
A nearly equiv. MS environment can be built using more hardware, plus lots of software licensing, and moderately skilled admin people.
For a small installation, MS may win on TCO, but as the data center gets larger and larger, it becomes cheaper to hire gifted admins rather than buy more servers/licenses - so the advantage swings to Linux.
At the desktop level, Linux runs well on older hardware. If the "Linux Terminal Server Project" style installation fits a substantial part of the business, the TCO can be staggeringly small. Again, it requires highly skilled admins. Linux can be tough on mobile users, people who constantly need to install differnet software, etc.
MS desktops need to be newer, they each need to be licensed, and since remote administration is not easy, lots of low skill techs are needed to run around helping users.
In the long run, Linux means you need less people but of better quality. In a larger environment that is offset by the reduction in hardware and licensing cost.
Why is it, that when someone describes in layman terms some basic exothermic chemistry, they are public enemy number 1? Should we hang the acedemics for teaching this chemistry? I am concerned about the wider scale of such generalized concepts in which people are categorized as criminals for learning and retaining knowledge that makes other's feel threatened.
Don't be silly. This guy is guilty because of two actions which, though individually are non-criminal, taken together become criminal. It is ok to have a discussion about making explosives. It is ok to advocate removing the government. Placing both discussions in the same context is not.
The reason he took the plea is that he didn't believe that he could convince even one of twelve people that his intent was not to promote murder and mayhem. Inciting violence is a crime, and that was clearly his intent.
Note very carefully that SCO targetted 2.4 based systems. IBM was not involved until many 2.4 kernels had already shipped. The alleged infringing code could not have been in all 2.4 kernels.
Thus SCO is attempting to license code that they themselves distributed under the GPL. This would invalidate the GPL, in which case SCO is guilty of copyright violation/software piracy for distributing Linus' and others' code.
Compilers don't always produce the same binary from the smae source.
Besides the obvious issues - which compiler/linker, precisely what version, what patches, which static libraries, what versions, etc - there is also the issue of alignment. I've worked with compilers/linkers that would not zero out empty space within the created images. Therefore the binary image would contain random gibberish that happened to be in memory when the compile ran. Thus, the "same" binary could generate different checksums.
Re:Spammers
on
FCC Goes WiFi
·
· Score: 3, Interesting
It seems pretty ironic that the FCC, while on the one hand leading a battle against spam, would choose to offer Wi-Fi, potentially catering to wannabe spammers.
I'm not sure that I understand this comment? I'm sure the FCC will not be connecting the access points directly to their internal network. A spammer sitting in their courtyard won't be able to access the FCC mail relay as if they were in the FCC. They'll only get "plain" access to the internet.
The spammers could run a sendmail process directly on their laptop that would route to the destination mail relay directly, but they could do this from any connection with any ISP. Does anyone filter outgoing SMTP packets? That would would be almost as bad as filter packets destined for port 80!
Loss Leader
on
FCC Goes WiFi
·
· Score: 4, Interesting
I've always subscribed to the theory that WiFi won't be profitable as a service itself, but will be provided for free as a "loss leader".
I always assumed however that it would be coffee shops and bookstores that used it to their competative advantage.
Next time I'm in DC, however, I'm going to patronize the FCC instead of any of the other federal agencies!
Seriously, Mike Powell seems to be on the forward edge, especially for a government official. It is good to have people in powerful places that understand technology and its transforming role, who can think independantly of lobbyists or position papers generated by their staff.
...the ease of use is only for USE. Not for support.
In certain business environments, Linux can be far easier to support. Many business users need a small suite of office and productivity apps, and not much more. A great setup is to put diskless workstations on each desktop, then run a few Linux terminal servers, locked down, in your datacom closet. Once the initial setup is done, maintenance is a breeze. Backups can be made from a central location. The user environment is portable to any workstation in the office. The admin can all be done in one place.
Obviously this is not a solution for every environment, but where is fits, it fits really well.
think the whole existance of this debate about A0 what type connector it actually is and B) what type would have been more typical proves that the design acheived the intended goal.
Yes, I confess to the count of silliness in the first degree.
Just my $0.10b cents worth.
What, why not an IEEE floating point number?
The recidivism rate is very high.
Patents, unlike copyrights, are enforceable against the user. The patent is on the process or the concept, not on any physical code. So an open source implementation, even from scratch, can be infringing.
Distributors and publishers can cover their asses by putting in a notice that the user may be required to license the patents.
Individuals are probably safe, since they are targets too small to hit.
Other companies will be the targets. IBM has enough patents to guaruntee mutual assured destruction, so they will be ignored/cross licensed.
Smaller companies may find themselves targets, but patents have a safety valve that they are re-examined by the court during litigation. Patents must be non-obvious to be enforcible, and many of the ones you listed won't meet that criteria in court, regardless of what the examiner decided.
In reality, the patent holder will likely either not attempt to enforce the patent, saving them for use against others who attempt to enforce other patents against them. Otherwise, patent holders will likely license the patents at low prices, making it cheaper for everyone to pay the license rather than litigate it.
Having a patent doesn't mean it can be effectively enforced.
I certainly may be wrong about the ATM part. I'm not an advanced networking guy.
The important point is that the protocol tunneled over the commodity IP connection, and was, by definition, able to use the entire bandwidth of the IP connection. The protocol supported QoS. It was implemented in software. Since the telco involved controlled both ends, it may have been a custom protocol that borrowed features from ATM.
My favorite is Chris Hoover's "Syringe becomes Modular plug".
But, being a geek, I have an annoying nitpick. The modular plug is for a three-pair cable (RJ-14). Why do a modem cable when broadband is so popular. Most ethernet connections are four-pair (RJ-45).
If you wanted to be retro geeky, use the BNC connector for 10Base-2 (ThinNet).
There's really no reason to pay a lawyer $150k/year.
You can't expect someone to spend three or four extra years in school, earn a doctorate degree, typically spend 80 hours a week in the office, and work for $30,000 a year. It just isn't going to happen. A top firm in a major city is going to start their attornies just out of school at 80k a year. By the time they are partners (8 years later) they're making $300,000+. Be realistic. A good sysadmin in the same markets frequently earns over $100,000 without graduate studies.
If you have twenty lawyers working full-time on these suits, plus paralegals, you need to get better lawyers. You should certainly be able to make do with many fewer per suit.
In many of these cases the discovery materials are supplies in truckloads. Hundreds of thousands of pages of documents need to be cataloged and reviewed, usually on a short time period like 60 days. Two or three lawyers aren't going to cut it.
..and all this is, is a glorified token ring implemented at the OS layer on top of 802.11.
Color me unimpressed.
Why are you critical of the solution? It appears to work well and is inexpensive. What is wrong with that?
A friend who used to work for a "baby bell" was involved with a project to provide VoIP and video services, as well as internet, over their DSL infrastructure. The problem that they ran into was that IP, as supported by their commodity DSL equipment, did not support QoS. Their solution was to tunnel a more advanced network protocol (ATM I think) over the entire DSL IP connection. Then they ran their voice and video over ATM directly and ran IP as another tunnel over ATM. It wasn't elegant, but it was cheaper and more effective than manufacturing new devices.
Anytime you can use commodity off-the-shelf hardware, you can usually save money.
Yes, I know he was critical of Linus on comp.os.minix, that is why I voted to create comp.os.linux. They are still excellent books.
This may make some sense...
on
Novell Buys Ximian
·
· Score: 5, Interesting
Novell is a name recognizable (and respectable) to the PHBs of the world. Sure they got trounced by MS, and their licensing structure may have sucked, but they are still a known name.
It would be easy for Novell to put together a nice bundle of Linux technologies, then sell it under their own name. The PHBs who don't trust OSS wouldn't have to know any better.
I'd personally like to see Novell hire the SAMBA team. It would be pretty cool to see them take back the file and print server space from MS using their name on OSS.
Even at $150,000.00 per year for every single person at a 20-person firm handling the case, and presuming they work on nothing else at the same time is still only $3,000,000.00. Add in say $1,000,000.00 (ridiculous) filing fees and direct court costs.
20 lawyer firm at $150,000 per year, plus benefits probably runs more like $4,000,000. Count in another 20 paralegals and clerks at $40,000 a year each, plus benefits comes to approximately $1,000,000. Office space, supplies, furniture, telephones, computers, yada-yada for 40 people and the reams and reams of documents, about $1,000,000. Advertising costs, which are required in a class action suit, $1,000,000. Taking your $1,000,000 for direct court costs, we are already at $8,000,000 and we haven't even hired an expert witness yet.
No other industry expects to reap a 1100% profit on a two year effort.
It is not unusual for a case of this size to cost $10,000,000. Furthermore most of these cases fail.
Therefore the economics are:
1) Bring four cases at a cost of $10,000,000 each.
the legal firms in such a case to charge full rate for the entire effort is obscene. Their profit on such cases should be capped at something reasonable, like 5-20% of their actual costs (filing fees, supporting research, etc. and not the lawyer's time. Their time is what the percentage is to cover, not double-dipping as both an hourly employee and as a profit-sharing partner of the firm.)
I have a great business deal for you.
You give me $10,000 and 1000 hours of work over the next 2 years. If the deal doesn't work out you get nothing. But if it does work out, I'll give you between $10,500 and $12,000 back.
Doesn't that sound great? It doesn't? Then why would you expect a lawyer to go for it? Your system means that no attorneys will take the risk, and so MS pays no penalty and the citizens of Florida don't even get their $12.
Slashdot Mirror
The Java language itself is fairly sparse. The collection of libraries that make up the JRE (Java Runtime Environment) is immense. And java is pretty useless until you learn your way around the APIs.
I like the language in general. The exception handling model is good. The standard APIs are rich. The namespace issues are handled well.
Developing in Java is a pain in the ass, however, due to classpath and packaging issues. The various IDEs help somewhat, but they can also obscure these issues and prevent you from learning them. In addition, everyone knows that vi is the one true editor - and vim provides good syntax highlighting too.
Best advice to a budding java developer is learn to use apache Ant right from the start. Using JUnit early on is also a good idea. Doing small projects with Ant and Java is far easier and more informative than doing projects in java alone or in an IDE.
Is it still has tedious to put the connectors on the ends?
When I was doing it, IIRC, the process ran something like this:
This should give those of you who have an old x86 PC that is too small to run even the smallest of Linux variants, a chance to browse the web, set up a web server, and doing other essential stuff.
If you have an old pre-386 machine around, why not run Minix? That should make a far more useful machine.
Actually, that is far from true. The details vary in different jurisdictions, but in general personal injury, medical malpractice, products liability, and class action suits like the one we're discussing are done on contigency. The lawyers must pay all expenses out of their own pockets, in addition to their labor. They can only receive payment if they win the case and collect the judgement.
I'm not sure how consistent this is in other jurisdictions and areas of law, but in NY State a personal injury or medical malpractice case by law pays one third of the judgement. It is illegal for an attorney to charge any other fee. In most of these cases, that 33% actually represents less than the normal hourly rate that an attorney would charge for other kinds of work.
My wife used to practice in this area when she worked for another law firm. When she went out on her own, she could not afford to risk the $10,000 - $100,000 from our own pocket to litigate a single case, a case that may take years to be completed.
I've got some ideas about TCO of windows vs. Linux, and I'm going to present them as facts.
In a corporate server environment, Linux can provide applications servers, database hosting, Web Server, and File & Print services through a suite of truly state of the art packages (JBoss, MySQL, Apache, Samba) for the cost of hardware and a few highly skilled admin people.
A nearly equiv. MS environment can be built using more hardware, plus lots of software licensing, and moderately skilled admin people.
For a small installation, MS may win on TCO, but as the data center gets larger and larger, it becomes cheaper to hire gifted admins rather than buy more servers/licenses - so the advantage swings to Linux.
At the desktop level, Linux runs well on older hardware. If the "Linux Terminal Server Project" style installation fits a substantial part of the business, the TCO can be staggeringly small. Again, it requires highly skilled admins. Linux can be tough on mobile users, people who constantly need to install differnet software, etc.
MS desktops need to be newer, they each need to be licensed, and since remote administration is not easy, lots of low skill techs are needed to run around helping users.
In the long run, Linux means you need less people but of better quality. In a larger environment that is offset by the reduction in hardware and licensing cost.
Why is it, that when someone describes in layman terms some basic exothermic chemistry, they are public enemy number 1? Should we hang the acedemics for teaching this chemistry? I am concerned about the wider scale of such generalized concepts in which people are categorized as criminals for learning and retaining knowledge that makes other's feel threatened.
Don't be silly. This guy is guilty because of two actions which, though individually are non-criminal, taken together become criminal. It is ok to have a discussion about making explosives. It is ok to advocate removing the government. Placing both discussions in the same context is not.
The reason he took the plea is that he didn't believe that he could convince even one of twelve people that his intent was not to promote murder and mayhem. Inciting violence is a crime, and that was clearly his intent.
Note very carefully that SCO targetted 2.4 based systems. IBM was not involved until many 2.4 kernels had already shipped. The alleged infringing code could not have been in all 2.4 kernels.
Thus SCO is attempting to license code that they themselves distributed under the GPL. This would invalidate the GPL, in which case SCO is guilty of copyright violation/software piracy for distributing Linus' and others' code.
Besides the obvious issues - which compiler/linker, precisely what version, what patches, which static libraries, what versions, etc - there is also the issue of alignment. I've worked with compilers/linkers that would not zero out empty space within the created images. Therefore the binary image would contain random gibberish that happened to be in memory when the compile ran. Thus, the "same" binary could generate different checksums.
I'm not sure that I understand this comment? I'm sure the FCC will not be connecting the access points directly to their internal network. A spammer sitting in their courtyard won't be able to access the FCC mail relay as if they were in the FCC. They'll only get "plain" access to the internet.
The spammers could run a sendmail process directly on their laptop that would route to the destination mail relay directly, but they could do this from any connection with any ISP. Does anyone filter outgoing SMTP packets? That would would be almost as bad as filter packets destined for port 80!
I always assumed however that it would be coffee shops and bookstores that used it to their competative advantage.
Next time I'm in DC, however, I'm going to patronize the FCC instead of any of the other federal agencies!
Seriously, Mike Powell seems to be on the forward edge, especially for a government official. It is good to have people in powerful places that understand technology and its transforming role, who can think independantly of lobbyists or position papers generated by their staff.
In certain business environments, Linux can be far easier to support. Many business users need a small suite of office and productivity apps, and not much more. A great setup is to put diskless workstations on each desktop, then run a few Linux terminal servers, locked down, in your datacom closet. Once the initial setup is done, maintenance is a breeze. Backups can be made from a central location. The user environment is portable to any workstation in the office. The admin can all be done in one place.
Obviously this is not a solution for every environment, but where is fits, it fits really well.
Yes, I confess to the count of silliness in the first degree.
Just my $0.10b cents worth.
What, why not an IEEE floating point number? The recidivism rate is very high.
Distributors and publishers can cover their asses by putting in a notice that the user may be required to license the patents.
Individuals are probably safe, since they are targets too small to hit.
Other companies will be the targets. IBM has enough patents to guaruntee mutual assured destruction, so they will be ignored/cross licensed.
Smaller companies may find themselves targets, but patents have a safety valve that they are re-examined by the court during litigation. Patents must be non-obvious to be enforcible, and many of the ones you listed won't meet that criteria in court, regardless of what the examiner decided.
In reality, the patent holder will likely either not attempt to enforce the patent, saving them for use against others who attempt to enforce other patents against them. Otherwise, patent holders will likely license the patents at low prices, making it cheaper for everyone to pay the license rather than litigate it.
Having a patent doesn't mean it can be effectively enforced.
C'mon, everyone. That was not an insightful post!
Actually, take a close look at the picture. It is an RJ-12. A reference page on phone connectors.
The important point is that the protocol tunneled over the commodity IP connection, and was, by definition, able to use the entire bandwidth of the IP connection. The protocol supported QoS. It was implemented in software. Since the telco involved controlled both ends, it may have been a custom protocol that borrowed features from ATM.
But, being a geek, I have an annoying nitpick. The modular plug is for a three-pair cable (RJ-14). Why do a modem cable when broadband is so popular. Most ethernet connections are four-pair (RJ-45).
If you wanted to be retro geeky, use the BNC connector for 10Base-2 (ThinNet).
You can't expect someone to spend three or four extra years in school, earn a doctorate degree, typically spend 80 hours a week in the office, and work for $30,000 a year. It just isn't going to happen. A top firm in a major city is going to start their attornies just out of school at 80k a year. By the time they are partners (8 years later) they're making $300,000+. Be realistic. A good sysadmin in the same markets frequently earns over $100,000 without graduate studies.
If you have twenty lawyers working full-time on these suits, plus paralegals, you need to get better lawyers. You should certainly be able to make do with many fewer per suit.
In many of these cases the discovery materials are supplies in truckloads. Hundreds of thousands of pages of documents need to be cataloged and reviewed, usually on a short time period like 60 days. Two or three lawyers aren't going to cut it.
Color me unimpressed.
Why are you critical of the solution? It appears to work well and is inexpensive. What is wrong with that?
A friend who used to work for a "baby bell" was involved with a project to provide VoIP and video services, as well as internet, over their DSL infrastructure. The problem that they ran into was that IP, as supported by their commodity DSL equipment, did not support QoS. Their solution was to tunnel a more advanced network protocol (ATM I think) over the entire DSL IP connection. Then they ran their voice and video over ATM directly and ran IP as another tunnel over ATM. It wasn't elegant, but it was cheaper and more effective than manufacturing new devices.
Anytime you can use commodity off-the-shelf hardware, you can usually save money.
Tannenbaum's best known work is Operating Systems, the Minix book.
Yes, I know he was critical of Linus on comp.os.minix, that is why I voted to create comp.os.linux. They are still excellent books.
It would be easy for Novell to put together a nice bundle of Linux technologies, then sell it under their own name. The PHBs who don't trust OSS wouldn't have to know any better.
I'd personally like to see Novell hire the SAMBA team. It would be pretty cool to see them take back the file and print server space from MS using their name on OSS.
Don't forget that there are a plethora of non-storage devices available for CF, including Wired and Wireless network cards.
20 lawyer firm at $150,000 per year, plus benefits probably runs more like $4,000,000. Count in another 20 paralegals and clerks at $40,000 a year each, plus benefits comes to approximately $1,000,000. Office space, supplies, furniture, telephones, computers, yada-yada for 40 people and the reams and reams of documents, about $1,000,000. Advertising costs, which are required in a class action suit, $1,000,000. Taking your $1,000,000 for direct court costs, we are already at $8,000,000 and we haven't even hired an expert witness yet.
No other industry expects to reap a 1100% profit on a two year effort.
It is not unusual for a case of this size to cost $10,000,000. Furthermore most of these cases fail.
Therefore the economics are:
1) Bring four cases at a cost of $10,000,000 each.
2) Lose three.
3) Win one to the tune of $48,000,000.
4) Profit $8,000,000 on a $40,000,000 outlay.
A %20 profit doesn't seem nearly as unresonable.
I have a great business deal for you.
You give me $10,000 and 1000 hours of work over the next 2 years. If the deal doesn't work out you get nothing. But if it does work out, I'll give you between $10,500 and $12,000 back.
Doesn't that sound great? It doesn't? Then why would you expect a lawyer to go for it? Your system means that no attorneys will take the risk, and so MS pays no penalty and the citizens of Florida don't even get their $12.