Except that you don't have to deploy it to user's desktops, there is never a client version incompatibility problem, doesn't require that users have Java 1.5.6.9, is faster than local Java applications both the load and run, and avoid MS lock in with.NET.
Right now we're migrating AWAY from a Java client to a browser, using AJAX to give a thick client feel. The familiarity of the browser and responsiveness of AJAX make Marketing happy, not have to release a maintain a version of client software makes Engineering happy. It works with our existing Linux servers, which also puts a smile on Engineering's face. Oh, yeah, its faster to develop a web application and tart it up with AJAX than to deal with Java. That makes Management happy.
Personal Quality Ranking, best-to-worst: Japanese, American, German.
My GM cars have all had good-to-great quality. My Chrysler was not good, and all three German cars lived at the shop. Price was no indicator of quality. My Mazda 626, which was $12k in 1991, never needed anything but scheduled maintenance. My 2001 BMW M5, which cost more than 6X the Mazda, spent more time in the shop than in my driveway.
Of the ten new cars that my household has purchased in the last 15 years, 3 were "Japanese", 4 were "American", and 3 were "German". The German cars were by far the worst, quality-wise. That is, they broke down and had to go to the shop. (A car isn't poor quality just because you don't like it. My Lumina was a crappy car, but it had great quality.)
Why did I use quotes around the nationalities? Let's put it this way: when I got my Accord, a neighbor ribbed me about buying a Japanese car rather than American. My response was, "My Accord was built in Youngstown, OH. Your Chevy trunk was built in Ontario. Which car is more 'American'?"
I worked at a dot-bomb through the latter half of 1999 and into the beginning of 2000. We were building an online drug store to complement the company's mail-order prescription drug renewal business. The idea was that you would use the web site to input your prescription number, and while at the web site, would purchase deodorant, some bandaids, and mouthwash. In short, the hope was for the online experience to mimic your experience in a brick-and-mortar pharmacy.
There was a real brick-and-mortar, mail-order prescription drug fullfillment business footing the bill for this. It had been started by a father. He was semi-retired and had turned the business over to his two sons. The Web site was their idea and they were in charge.
We had a million dollars in middleware, a couple million in consulting to customize the middleware, an Orcale backend running on a high end Sun (E7500), and the Web site itself running on a top of the line, Sun E10k. At this point there was about $5 million sunk into the project, and we had not yet gone live.
Before going live, management felt the need to run a load test. At that point, you saw the IBM commercials on TV were dot coms went live only to see the site crash due to too much traffic. They didn't want to see that happen. The load tests showed that we could only handle 1000 simultaneous transactions. Clearly, that wasn't enough. So we bought another E7500, another loaded E10k, and another Oracle license. I don't know the exact numbers but I think this was close to another $3 million. With this new equipment and an additional DS3 line, we could handle 2500 simultaneous transactions.
Early in 2000 it comes time to turn the web site live and crank up the advertising. Tension was running high - and expectations were greatly disappointed. The largest number of visitors we ever had to the site was eight. We never had more than one active transaction.
I only stayed around for another couple of months. Before I left, the father, who founded the business and ultimately footed the nearly $10 million dollar tab, said:
It would have been cheaper to have phone operators and advertise that we would wrap every package in $100 bills.
This is the real thing, with the optimizing compiler and the IDE. The Express editions lack the OS header files (windows.h, etc). This means you cannot access network, file systems, etc. I'm pretty sure that the Express editions also lack the very good optimizing compiler, but its been a while since I used Express.
Using an old, and not very smart, C++ compiler (gcc 2.95 plus vendor patches) we recently measured the cost of throwing an exception as 8-10 cycles. This was an empty class, to avoid measuring construction overhead e.g.
class Exception {};
That doesn't strike me as an "obscene" performance hit.
I had a '98 Corvette convertible (stock) and regularly got 30+ mpg on the highway. Around town I typically got 18-20mpg. My friend has an '04 Z06 and gets 28+ mpg on the highway.
If it were the trial lawyers, once states enact draconian medical liability reform then insurance rates should come down. We've done this in Texas, and doctor's insurance rates are not coming down.
Similarly, home owners insurance rates went up an average of 70% three years ago, supposedly for black mold coverage. The Leg. retroactively removed coverage for black mold and water damage two and a half years ago. My homeowner's insurance rates have yet to go down.
Trail lawyers are a scapegoat propagated by business interests so they can increase profits (or offset losses incurred by bad.com investments).
Everyone likes to blame "those damn lawyers" until they need one. Businesses blame a large number of "frivolous" lawsuits on increasing their legal bills. Who defines frivolous? Businesses still initiate 80% of legal actions in this country. And businesses are not hesitant to unleash lawyers on indivuals.
IANAL, I am a design engineer. But I'm tired of hearing Chamber of Commerce misinformation quoted like its unbiased. Lawyers look after their clients interests. Lawyers for businesses look out for the business. If you have a dispute and no lawyer, who is looking out for your interests?
Where can I find this XFree86 integrated with Windows?
I've been using the doesn't-work-with-Gnome and crashes-under-Emacs Cygwin version. We use it at work to connect to the Linux servers running RedHat AS. Due to limitations and stability problems with Cygwin XFree86, we're now looking at VNC solutions.
So if there is a good, free X server for Windows, I'd like to know where to find it.
I used one of those too, only it was in a computer class at a summer camp. Output was one row, 8 "digits". I was about 12, and the instructor put a lot of numbers on the board for us to load (via switches) into the machine. The we hit the execute button and it said, "What are you looking at froggy-face?"
When I was an undergraduate, I presented at an image processing/image understanding conference at the U. of Texas in Austin. Someone presented a paper the idea of using mountains or buildings to determine position. There was a demo program done on an SGI, where a long distance rover could recognize its position from Martian topography.
Once I bought a PC (yes, I bought it.) and it kept crashing after running NT for 30-60 minutes. I couldn't convince the vendor that there was a problem until I tried to boot linux and it would kernel trap during boot complaining about bad memory.
After demonstrating the (reproducible) problem the vendor replaced the second SIMM and all was well.
In my lab we're working on surgical simulation including organ physics, cutting, bleeding, etc. We need all the perfromance we can possibly get. The GPU is a monsterously fast parallel vector processing engine, and can be used for non-graphics computation. Asymmetric AGP bandwidth has prevented us from using the GPU as a coprocessor thus far: across the AGP bus you can push data down to the GPU through a firehose, back up through a straw.
AGP was a hack onto PCI. PCI-Express will give us the symmetric bandwidth we need. Yeah!
Configuration Management is about a lot more than version control. It involves creating a branching-merging process that can support the development model. CM is responsible to know exactly what went into a build, verify that it reproducible, and then control distribution pending approvals, e.g. testers don't get the build until development management approves, customer's don't get the build without project management approval.
These processes are automated so that they can be done repeatable. A variety of reports are generated, including things like "Did all the files with a 'buffer overrun branch get merged for this build?"
Clearcase is a great tool. Its real value to me is that by being a ClearCase administrator, it elevates me above the "I have SourceSafe experience" crowd.
Finally, ClearCase is complex because it is powerful and flexibile. Large development groups are composed of developers with different levels of maturity and experience. Software from hundreds of developmers needs to be combined into a deliverable with known composition that is reproducible.
No software package will replace me in the forseeable future, although large development projects going overseas has cut back my work.
I don't know what you consider "scaling up" but I've administered ClearCase with replication across 3 continents, 7 time zones, and 500 developers. Granted, each physical site has its own server. But the main site had almost 200 people using Clearcase, the majority of them developers who were hitting it minute by minute.
Considered ownership rules and triggers are necessary, but it works better than anything else I've seen for multisite development.
Its not true that Clearcase is one huge table internally. I do database repairs on Clearcase, and there are certainly multiple tables. Careless queries can generate a HUGE amount of data, but this is true of any large DB.
SCM is how I earn my living. I install, maintain, and development SCM tools, processes, and automation. While I haven't used BitKeeper, I've worked on a whole raft-load of others. ClearCase, while monsterous in its complexity, is the head-and-shoulders winner for large scale development.
ClearCase is a version database with a powerful query language. You can select file versions by label, date, owner, attributes, branch, etc. Because ClearCase is its own filesystem, it can (but doesn't have to) transparently update your view to the latest versions. (This feature is nice for server rollouts. If something is wrong, change the query and you're instantly back to the old.) Also because it is its own filesystem, ClearCase provides audit builds that tell you everything that was touched when creating a derived object.
Updates for your view are lightning fast. And it can version directory trees, which is sorely missing in most other products. It has a documented Perl API.
On the downside ClearCase is roughly as difficult to administer as Oracle. It is expensive in terms of dollars, server hardware, and network resources.
That's it. No publishing your source.
We did some tests of various compression algorithms for compressing executables. (Granted, not the same as text.) We tested compress, zlib (via gzip), bzip2, pkzip, and a few others. bzip2 produced the best compression by 30% but took 2.5X. zlib produced the best decompression speed/compression ratio.
I was sued by a small corporation that I used to work for. After two years and $135,000 I won. Yeah, I won but I'm still out more than the value of my house. And that was a small (~10 people) firm without staff attorney's like the RIAA.
My business is software configuration management. Electronic software delivery is a critical part of many solutions. Typically we use a web site. The system has access control, software submital, notification, approvals at various levels, retrieval based on approval level, and logging.
For examply, only users identified as Development can submit software. At that point Software Configuration Management is notified to reproduce the software (can SCM build the same binaries as the developers?) SCM retrieves the software from the web site. Once SCM approves the software, Test is notified.
Test retrieves the software and puts it through its paces. If it passes Test grants its approval through the web site. Otherwise the software fails and Test provides a URL explaining the problems. And on...
At any point program management can see the state of the software in its track to customer delivery. PM has override ability to approve software for customer delivery even if it has, for example, failed testing.
The web site makes it easy to access. Access control and approval manage the software delivery process. Notification keeps everyone on the ball. And logging provides CYA - and has covered my butt on numerous occasions.
My boss particularly loves to be sitting in a Change Control Meeting and hear the development manager say, "The software's been delivered to SCM. We're waiting on them." And he can say with confidence, "Not yet it hasn't."
Here in central Fort Worth (700,000 strong), within walking distance of a University (Texas Christian University) we have only two choices: dial-up or Charter cable modem. DSL is NOT available in this area, despite being within 4 miles of downtown.
Charter has consistently downgraded serivce in the three years its been available, with two steps-down in speed (3Mbps -> 1 Mbs -> 128 kbps), changing from static IPs to DHCP, and going from unrestricted to port blocking (no mail servers, web servers, etc.)
If they offered a higher class of service (static IP, ability to run servers are important to me, 128 kbps isn't a big problem) I'd jump on it. They keep talking about adding better service tiers, but never get around to it.
Bell Labs released vmalloc() for public use and gave a white paper on it at 1996 USENIX. I recently investigated it for memory leak problems in an embedded real time system.
http://www.research.att.com/sw/tools/vmalloc/
Slashdot Mirror
Except that you don't have to deploy it to user's desktops, there is never a client version incompatibility problem, doesn't require that users have Java 1.5.6.9, is faster than local Java applications both the load and run, and avoid MS lock in with .NET.
Right now we're migrating AWAY from a Java client to a browser, using AJAX to give a thick client feel. The familiarity of the browser and responsiveness of AJAX make Marketing happy, not have to release a maintain a version of client software makes Engineering happy. It works with our existing Linux servers, which also puts a smile on Engineering's face. Oh, yeah, its faster to develop a web application and tart it up with AJAX than to deal with Java. That makes Management happy.
UPS, of the brown vans, has a service mark on "Brown." While sitting on an airline, a friend and I were talking about how ridiculous a copyright on silence was. http://legalminds.lp.findlaw.com/list/cyberia-l/ms g41561.html
At that point, I flipped the magazine page in my lap and saw a UPS ad. At the bottom it showed " Brown(SM) ". Yikes!
Personal Quality Ranking, best-to-worst: Japanese, American, German.
My GM cars have all had good-to-great quality. My Chrysler was not good, and all three German cars lived at the shop. Price was no indicator of quality. My Mazda 626, which was $12k in 1991, never needed anything but scheduled maintenance. My 2001 BMW M5, which cost more than 6X the Mazda, spent more time in the shop than in my driveway.
Of the ten new cars that my household has purchased in the last 15 years, 3 were "Japanese", 4 were "American", and 3 were "German". The German cars were by far the worst, quality-wise. That is, they broke down and had to go to the shop. (A car isn't poor quality just because you don't like it. My Lumina was a crappy car, but it had great quality.)
Why did I use quotes around the nationalities? Let's put it this way: when I got my Accord, a neighbor ribbed me about buying a Japanese car rather than American. My response was, "My Accord was built in Youngstown, OH. Your Chevy trunk was built in Ontario. Which car is more 'American'?"
There was a real brick-and-mortar, mail-order prescription drug fullfillment business footing the bill for this. It had been started by a father. He was semi-retired and had turned the business over to his two sons. The Web site was their idea and they were in charge.
We had a million dollars in middleware, a couple million in consulting to customize the middleware, an Orcale backend running on a high end Sun (E7500), and the Web site itself running on a top of the line, Sun E10k. At this point there was about $5 million sunk into the project, and we had not yet gone live.
Before going live, management felt the need to run a load test. At that point, you saw the IBM commercials on TV were dot coms went live only to see the site crash due to too much traffic. They didn't want to see that happen. The load tests showed that we could only handle 1000 simultaneous transactions. Clearly, that wasn't enough. So we bought another E7500, another loaded E10k, and another Oracle license. I don't know the exact numbers but I think this was close to another $3 million. With this new equipment and an additional DS3 line, we could handle 2500 simultaneous transactions.
Early in 2000 it comes time to turn the web site live and crank up the advertising. Tension was running high - and expectations were greatly disappointed. The largest number of visitors we ever had to the site was eight. We never had more than one active transaction.
I only stayed around for another couple of months. Before I left, the father, who founded the business and ultimately footed the nearly $10 million dollar tab, said:
I went to this link http://msdn.microsoft.com/getthebetas/ and gave MS $12 and they shipped me two copies of VS2005 Beta 2. They arrived in two days.
This is the real thing, with the optimizing compiler and the IDE. The Express editions lack the OS header files (windows.h, etc). This means you cannot access network, file systems, etc. I'm pretty sure that the Express editions also lack the very good optimizing compiler, but its been a while since I used Express.
That doesn't strike me as an "obscene" performance hit.
I had a '98 Corvette convertible (stock) and regularly got 30+ mpg on the highway. Around town I typically got 18-20mpg. My friend has an '04 Z06 and gets 28+ mpg on the highway.
Similarly, home owners insurance rates went up an average of 70% three years ago, supposedly for black mold coverage. The Leg. retroactively removed coverage for black mold and water damage two and a half years ago. My homeowner's insurance rates have yet to go down.
Trail lawyers are a scapegoat propagated by business interests so they can increase profits (or offset losses incurred by bad .com investments).
Everyone likes to blame "those damn lawyers" until they need one. Businesses blame a large number of "frivolous" lawsuits on increasing their legal bills. Who defines frivolous? Businesses still initiate 80% of legal actions in this country. And businesses are not hesitant to unleash lawyers on indivuals.
IANAL, I am a design engineer. But I'm tired of hearing Chamber of Commerce misinformation quoted like its unbiased. Lawyers look after their clients interests. Lawyers for businesses look out for the business. If you have a dispute and no lawyer, who is looking out for your interests?
I've been using the doesn't-work-with-Gnome and crashes-under-Emacs Cygwin version. We use it at work to connect to the Linux servers running RedHat AS. Due to limitations and stability problems with Cygwin XFree86, we're now looking at VNC solutions.
So if there is a good, free X server for Windows, I'd like to know where to find it.
I used one of those too, only it was in a computer class at a summer camp. Output was one row, 8 "digits". I was about 12, and the instructor put a lot of numbers on the board for us to load (via switches) into the machine. The we hit the execute button and it said, "What are you looking at froggy-face?"
When I was an undergraduate, I presented at an image processing/image understanding conference at the U. of Texas in Austin. Someone presented a paper the idea of using mountains or buildings to determine position. There was a demo program done on an SGI, where a long distance rover could recognize its position from Martian topography.
SunOS had its roots in BSD. Solaris has it roots in SVR4. The changeover happened in the '94-'95 time frame.
After demonstrating the (reproducible) problem the vendor replaced the second SIMM and all was well.
AGP was a hack onto PCI. PCI-Express will give us the symmetric bandwidth we need. Yeah!
Actually, our dorm was warned once by the vending company, he kept doing it and got caught on the security camera.
These processes are automated so that they can be done repeatable. A variety of reports are generated, including things like "Did all the files with a 'buffer overrun branch get merged for this build?"
Clearcase is a great tool. Its real value to me is that by being a ClearCase administrator, it elevates me above the "I have SourceSafe experience" crowd.
Finally, ClearCase is complex because it is powerful and flexibile. Large development groups are composed of developers with different levels of maturity and experience. Software from hundreds of developmers needs to be combined into a deliverable with known composition that is reproducible.
No software package will replace me in the forseeable future, although large development projects going overseas has cut back my work.
Considered ownership rules and triggers are necessary, but it works better than anything else I've seen for multisite development.
Its not true that Clearcase is one huge table internally. I do database repairs on Clearcase, and there are certainly multiple tables. Careless queries can generate a HUGE amount of data, but this is true of any large DB.
ClearCase is a version database with a powerful query language. You can select file versions by label, date, owner, attributes, branch, etc. Because ClearCase is its own filesystem, it can (but doesn't have to) transparently update your view to the latest versions. (This feature is nice for server rollouts. If something is wrong, change the query and you're instantly back to the old.) Also because it is its own filesystem, ClearCase provides audit builds that tell you everything that was touched when creating a derived object.
Updates for your view are lightning fast. And it can version directory trees, which is sorely missing in most other products. It has a documented Perl API.
On the downside ClearCase is roughly as difficult to administer as Oracle. It is expensive in terms of dollars, server hardware, and network resources.
- Not claim you wrote zlib
- Plainly mark altered version s of the source
- Not remove copyright header from source.
That's it. No publishing your source.We did some tests of various compression algorithms for compressing executables. (Granted, not the same as text.) We tested compress, zlib (via gzip), bzip2, pkzip, and a few others. bzip2 produced the best compression by 30% but took 2.5X. zlib produced the best decompression speed/compression ratio.
You can win a lawsuit and still lose big time.
I was sued by a small corporation that I used to work for. After two years and $135,000 I won. Yeah, I won but I'm still out more than the value of my house. And that was a small (~10 people) firm without staff attorney's like the RIAA.
My business is software configuration management. Electronic software delivery is a critical part of many solutions. Typically we use a web site. The system has access control, software submital, notification, approvals at various levels, retrieval based on approval level, and logging.
For examply, only users identified as Development can submit software. At that point Software Configuration Management is notified to reproduce the software (can SCM build the same binaries as the developers?) SCM retrieves the software from the web site. Once SCM approves the software, Test is notified.
Test retrieves the software and puts it through its paces. If it passes Test grants its approval through the web site. Otherwise the software fails and Test provides a URL explaining the problems. And on...
At any point program management can see the state of the software in its track to customer delivery. PM has override ability to approve software for customer delivery even if it has, for example, failed testing.
The web site makes it easy to access. Access control and approval manage the software delivery process. Notification keeps everyone on the ball. And logging provides CYA - and has covered my butt on numerous occasions.
My boss particularly loves to be sitting in a Change Control Meeting and hear the development manager say, "The software's been delivered to SCM. We're waiting on them." And he can say with confidence, "Not yet it hasn't."
The tiny 6 pt fixed font on this page makes it unreadable. Not everyone is 19 or has strong eyes.
In the future lets pick review sites that allow the Web client to format the fonts.
Here in central Fort Worth (700,000 strong), within walking distance of a University (Texas Christian University) we have only two choices: dial-up or Charter cable modem. DSL is NOT available in this area, despite being within 4 miles of downtown. Charter has consistently downgraded serivce in the three years its been available, with two steps-down in speed (3Mbps -> 1 Mbs -> 128 kbps), changing from static IPs to DHCP, and going from unrestricted to port blocking (no mail servers, web servers, etc.) If they offered a higher class of service (static IP, ability to run servers are important to me, 128 kbps isn't a big problem) I'd jump on it. They keep talking about adding better service tiers, but never get around to it.
Bell Labs released vmalloc() for public use and gave a white paper on it at 1996 USENIX. I recently investigated it for memory leak problems in an embedded real time system. http://www.research.att.com/sw/tools/vmalloc/