ALS - when it was the ATLANTA Linux Showcase - was a techie's dream. A small show with lots of Birds of a Feather sessions and immediate access to the folks actually doing things. I attended the past three years. Last year I got to go to the OpenSource Conference in Monterey but found ALS more enjoyable.
But I won't travel to Oakland for ALS. There are too many other shows competing. In Atlanta they were the only show around and the quality showed. Atlanta is also a day-trip for me whereas Oakland requires effort.
Right now, HIPPA guidelines can be met with 128-bit SSL and user id/passwords. In the not-too-distant future, HIPPA will begin require 'non-repudiated' authentication. This means the use of X.509 browser certificates for access to medical data and S/MIME for emails (OK, you could use PGPmail and some kind PGP Java authentication system, but that's quite unwieldy).
While a small medical clinic may be able to setup a Certificate Authority (CA), you may need to use Verisign or another large CA to issue these certificates. But you're biggest stumbling block is user support. Installing browser certificates and educating the user on the safe use of certificates is quite a challenge.
When I heard that Sony was shipping 1/2 as many PS2s as planned for X-Mas I decided that I was going to have to buy a DVD player and stop watching movies on the computer. I did try one of the X10 DVD-Anywhere kits which was interesting, but...
So I started looking at players planning to spend about $150 and noticed that you can't get DTS at that price point. I started browsing the auctions and found an Onkyo DR90 at UBID.com for $349. It actually is a DTS Receiver/Amp with the DVD player built-in. I wanted a set of 5.1 speakers to go with it and UBID was selling the Warfedale Movie Modus 5 system for about $149. I'd never heard of Warfedale before but I figured the price was right. I was quite pleased with what I got. The Onkyo DR90 automatically selects the best settings for the DVD you are playing and the Warfedales are at least competent. The system may not be earth-shattering but for less than $500 I've been quite happy.
NOTE: The one feature this system lacks that will cause me to replace it all in a year or two is progressive scan. I never noticed how bad the interlacing problems were on DVD when I was playing movies on my computer!
Gee, I have a single config tool: it's called vi (on a minimal system).
Microsoft really f*cked up windows for many of us when they moved away from *.ini to the registry. Personally, I think flat-files for configuration are the pinnacle of configuration methods. XML might seem equally easy to edit, but the parsers would need to preserve formatting and comments. Why not just standardize on
# this is a comment
keyword=setting
We're almost there already and there are plenty of libraries out there to read and write them.
I guess the only real advantage of XML conf files would be that your GUI config utility wouldn't have to know about a new tool before it could give you the config options for it. But my philosophy is that if you have to use a GUI to config your system, maybe you shouldn't be config'ing it.
A few years back I was writing software to control tufting machines (like big sewing machines for making carpet). I almost used the GameBoy as the platform because it took almost zero processing power (the GameBoy has an 8Mhz Z80 clone - which is quite reasonable) and needed only about 64K to run, including the code and buffers to display a graphic of the patter. The GameBoy can actually address about 16MB of memory via page-flipping.
The real benefit of the GameBoy is if you try to price an industrial control with a few buttons and a nice LCD, you quickly exceed the $50 in small quantities that a GameBoy at Toys'R'Us costs.
Our HP NetServer LH6000's are running like a top. They did come with the wrong RAID BIOS, so Red Hat wouldn't recognize the RAID drives. Tech Support was very efficient and pointed us right to the problem. These are $40K+ boxes (6-way Xeon, 1GB RAM, 6x18GB RAID5). We are running Sybase 11.0.3 (about to move to 11.9.2) on them.
HP seems committed to providing the same level of support for Linux on Intel as they have for HPUX on PA-RISC. They are even offering Linux-oriented training courses through their educational programs.
Of course, supply on some parts, like the Xeons, is variable and the delivery time for one of these servers is way slow. And don't even think of running Mandrake (or really any distribution other than Red Hat) on the big boxes. The hardware support just isn't there.
We also have a couple HP LPr's, which have been selling on Onsale for cheap.
For smaller stuff, we have a slew of Penguin Computing 1RU servers. We order whatever Penguin has in stock because the delivery time is everything.
We also are using the Penguin Computing dual-head workstations for development systems. I'm not thrilled with how the Matrox G400 stretches a single desktop across both monitors (my dialogs pop up in the crack in between) but who am I to complain. Penguin also does a very good with their preinstalled distribution on their workstations. Upgraded kernel, Xi Graphics X Server, etc., etc.
Actually I would like to see a Samba port to NT. Samba provides much better control over Browse Master and PDC election than Windows SMB Server and it would be cool to be able to configure the thing with a text file.
Of course, I don't want to see it bad enough to do the code myself.;)
Re:This is no longer the case with me.
on
Interface Zen
·
· Score: 1
Hmmm... My IBM 101Key keyboard I'm clacking away on right now is dated 11-NOV-91 (on the bottom). It's one of the newer one's I own. I started using these keyboards exclusively about five years ago. I used to try to just use whatever I was given, but once I started with the IBM keyboard I saw my accuracy and productivity go up. IBM has been making keyboards since before there were computers. The also made some of the first keyboards that didn't drive mechanical systems (i.e., electric typewriters). They really know what there are doing...
As far as keys in the penalty zones, I use CRiSP and find that my hands work in two modes: navigation/editting and creation. I guess this is analogous to edit and insert modes in vi. In navigation/edit mode my left hand hovers over the left shift key and my right hand hovers over the arrow keys. In edit mode my hands land on the home row.
CRiSP (like pretty much all editors I've used) provide ways to use the arrow keys to move more than one character. Ever try holding down the CTRL key while left or right arrowing? It even works in web browsers! And Home and End work pretty good too.
I really think the world is divided into two types: vi users and non-vi users. Personally I think vi is a single generation away from edlin and it's ilk. Give me windows, cut & paste, and modelessness, or give me death...
BTW, I use CRiSP in text mode more than I do in X windows. The beauty of CRiSP is all of the keystroke commands work identically in either CUI or GUI.
There are many languages that aren't inherently based on English, and more importantly, approach programming for very different points of view. Ever try forth? C++ provides operator and function overloading that could be used to create a non-English sublanguage.
But that's not the problem. Computers are procedural devices. Or rather, computers based on current CPU design are procedural. The processor executes one instruction, and then the next, and then the next. There are certain operations that this model requires: loops, branches, etc. Whether you call them "while", "for", "djnz", or something else, you still have the same basic concept. Even SMP and Massively Parallel systems are procedural - they just do more than one thing at once.
This is not an English thing. Changing the names of the commands won't change the way we write programs.
If you want to play around with a truly different programming paradigm, try neural networks.
And another note: I've worked with non-English speaking programmers. They tend to misspell variable names, but do it consistently!
Hmmm. I think are work (McKee Foods) I'd have problems with those names. First of all, we are forbidden to even mention Twinkies or Ho-hos. Second, if we named a machine Swissroll or NuttyBuddy or StarCrunch we'd confuse it for one of the production lines...
Instead we have really lame-o names like ORAD1 (Oracle Development 1), MFG1 (Manufacturing 1), etc...
Umm... Actually, the Air Force won't let you pilot a plane unless you have perfect 20/20 or better in both eyes uncorrected. I went through the academy physical in high school and they had to dilate my pupils for the vision test. I'm slightly off 20/20 in my left eye (couldn't even be labeled as 25/20, just NOT 20/20). It was enough to keep me from ever being able to fly.
Not only that, but you have to retest your senior year in the academy and still have 20/20 or better. Can you imagine the strain of four years of CompSci at the Air Force academy?
The Army has similar restrictions for helicopter pilots as does the Navy for Hydrofoil captains.
There are also many other restrictions for flying: height, weight (of course), you can never have blacked out in your life, and on and on...
Just kidding. I've been doing the Oracle thing for almost a year and am still astounded at just how bad SQL*Plus is. Unfortunately I use it everday. Most of my development is in C using Pro*C, so GUI tools aren't much help.
I am about to setup my own web server with an Oracle backend. I want to play around with WebDB and PHP3. I also need to play around with Developer because Oracle doesn't offer a certification program for 3GL programmers and I'm not ready to get into Oracle's Java mess.
BTW, PL/SQL, the Oracle procedural language, actually beat out Perl in my book as the most obscure language. The syntax is really funky and the use of semicolons is inconsistent with every language I've ever used.
But enough griping. If Oracle didn't make such screwy interfaces, people like me wouldn't make as much money making them work...
Remember that old GEOS operating system from the C64/C128? It grew into GEOS on PC in the late 80s, early 90s (the original AOL software used it!). It ran great on a 8088 with 640K RAM. It was an easy-to-use GUI that was legendary in it's robustness.
Guess what, it still exists:
http://www.newdealinc.com
It's grown up a little bit - best run on a 386 with at least 1MB of RAM, but still works on 286s. It includes a complete office suite and a web browser. There's even a BASIC interpreter.
The neatest feature it has is in the control panel you can set the user's skill level on a 1-4 range. All New Deal apps follow this setting.
A friend of mine who runs an ISP has New Deal on a 486 that his adopted daughter with Downs Syndrome uses. She's able to draw in the paint program, play solitaire (actually, she just moves the cards around), types in the word processor and browses the web.
It really doesn't get much easier. And as far as administration goes - anything that would kill New Deal would also take down Linux.
And I think I should try to find another forum for this discussion. This is a VERY important topic for many Oracle folks...
My big question about this post is:
When you say it blows chunks, are you saying SQL*Plus blows chunks or does the server instance actually give up the ghost? I'll have to try it here at work (I think we've got an 8i instance around here someplace...)
Not quite. You can download RPMs of ASE 11.0.3 for development and evaluation purposes, but when you actually implement an application with it the license fees are the same as the NT version - about $300/user or a flat $50K if the app touches the web.
Oracle's fees are similar but they are pushing use of "application service providers". Basically web hosts that have licensed Oracle instances that you can use for a premium.
BTW, if you want a free set of Oracle Development CDs, check out:
Register for the Atlanta or Birmingham seminars. They are fully booked - you won't get to go. But Oracle will send you a complimentary set of CDs for free
I've used PostgreSQL and Sybase for a few Linux based projects. I use Oracle 7.3 on HPUX at work. I'm helping a friend spec out a new project. He's ultimately going to use Oracle on Solaris (its government work - so the DB & OS choice is decided through politics). But development will likely occur on Oracle on Linux.
We have been going around in circles trying to decide whether to use Oracle 8.0.5 or 8i. As far as I can tell, the only significant advantage to 8i is the JVM. But 8i requires a minimum of 128MB RAM and 256MB swap without the JVM and 256MB RAM and 512MB with the JVM. Note: those are the minimums. And Oracle's not kidding. Don't even try firing up an 8i instance with less than 128MB RAM. Ultimately, the memory requirements will be satisfied. I suspect we'll spec out a Sun Ultra20 with about 1GB of RAM. But We don't want to invest a fortune in the development hardware.
Oracle 8.0.5, on the other hand, runs nicely with at little as 32MB of RAM. Plus, the install is CUI - whereas 8i requires X (I usually don't install X on my servers).
Besides a few cool new features in PL/SQL, does anyone else have any experience with 8.0.5 vs. 8i that would help sway me one way or the other? Right now we are pretty much set on using 8.0.5...
No RAW IO on Linux. I won't go into detail here, but Linux doesn't support Raw IO right now. Linus himself has stated that he has no interest in adding raw IO support because the performance increase from using buffered IO far exceeds the disaster recovery benefits of raw IO. But I hadn't though about FS limits. That's not so much a problem with Linux or Intel, but rather a problem with the POSIX file manipulation functions in GLIBC or LIBC. What does ftell return? A signed 32-bit long, which has a range of -2GB to +2GB. I think there are some kludges that use an unsigned long, but until 64 bit POSIX functions, we're stuck.
To a certain degree, Microsoft does provide developer contact. They just don't publish it in "mainstream" press.
The Microsoft Developer's Network is the first line of contact with the developers: "here's a huge stack of CDs of everything we think you might need to develop for our OS."
You can also get significant interaction with developers and evangelists at trade shows. When I was programming games I found that the time spent at Microsoft Meltdown was invaluable. You get to schedule time in a hotel suite with DirectX developers from Microsoft to review your code and get pointers on how the OS works. Plus, it's their front-line for input on new features to add.
I was even invited to spent a few days in Redmond at the offices to get direct interaction with DirectX developers.
However, Microsoft is a traditional company when it comes to trade secrets. You didn't get invited to Meltdown without signing NDAs. OpenSource eliminates the need for NDAs and lawyers in general.
Microsoft's position is also bound by the fact that they are in the business of creating de facto standards. Linux is based on open standards. The concept of open community interaction is there from the initial stages of development.
My friend (who is reading over my shoulder) and I have been through two different game companies (actually three if you consider the first company before the buyout). While I've never enjoyed a job more, I don't think I'd want to get back into the market. The game market has really closed up a great deal compared to 5-10 years ago.
It's almost impossible for a new shop to take off. You pretty much have to develop a game completely before the distributors take you seriously and the game has to compare to what established shops were producing.
For instance, the first company I worked for made statistically accurate baseball and football games exclusively for the PC. Our boss wanted to go head-to-head with EA and try to add arcade style play and graphics. Well, the investment to make that leap is huge. We tried to do it with a team of 15 people - about 5 programmers and the rest artists and sports experts. We tried to turn out three titles a year - baseball, pro football and college football. We didn't even come close. I later interviewed with EA's division that does the Madden NFL games and found out that they have over 20 people working just on the pro football game alone.
Not to mention the disheartening sight of a streatch Hummer at E3 custom made for EA sports. That damned truck cost more than the production budget of our last title.
I too have moved on to ecommerce development and have experienced all of the benefits that are listed above. Better pay, reasonable hours, female coworkers, and on, and on.
I have also been asked dozens of times what it would take for someone to become a game developer. I always tell them - don't. It just isn't the career it used to be.
It is just the render engines. Pixar has a port of Renderman for Linux too (and they are working on the Alpha Linux port as we speak). And, of course, the Blue Moon Render Tools have existed on Linux for years. BMRT implements a subset of Renderman - I've messed around with exporting the geometries out of Alias on Irix and rendering on BMRT on Linux but didn't have enough management support to get beyond the initial test.
Alias released the renderer for NT about 3-4 months before the full Maya toolset. It was worth the wait. Of course, I'm more interested in seeing the good bits of Irix move to Linux. The release of SGI Linux 1.0 was more significant to me. If SGI put 1/10th the effort into Linux that they did Irix, NT would die quickly.
Ummm... If I remember right, Neural Nets are essential parallel-style nodes with extremely limited intelligence. It is the density of the interconnections between the nodes that actually create the "intelligence". Multiprocessor parallel systems are usually based on very intelligent nodes with limited interconnections (i.e., each node has a big Pentium or Alpha with local memory) and only one network connection.
Parallel Processing is like a think-tank. Lots of very bright people working on individual problems. Communication and coordination of the individuals becomes a major task and any problem submitted has to be diced up into individually bite-sized morsels.
Neural Nets is like a brain. A bunch of independently dumb cells with lots of interconnections. Communcation between the cells occurs "naturally" and the problem flows through the cells like an amorphous blob.
There is no easy way to take advantage of parallel processing. It has to be designed for early in the development process. What this library proposes is to write all applications for distributed processing, even if the software is normally run on single CPU or SMP boxes.
Of course, there are plenty of problems that are inherently easy to implement in parallel, brute-force crypto-cracking, scanning sections of data for coherent signals, packetized compression, graphics rendering, etc. Heck, even make can be parallelized easily.
Slashdot Mirror
ALS - when it was the ATLANTA Linux Showcase - was a techie's dream. A small show with lots of Birds of a Feather sessions and immediate access to the folks actually doing things. I attended the past three years. Last year I got to go to the OpenSource Conference in Monterey but found ALS more enjoyable.
But I won't travel to Oakland for ALS. There are too many other shows competing. In Atlanta they were the only show around and the quality showed. Atlanta is also a day-trip for me whereas Oakland requires effort.
Right now, HIPPA guidelines can be met with 128-bit SSL and user id/passwords. In the not-too-distant future, HIPPA will begin require 'non-repudiated' authentication. This means the use of X.509 browser certificates for access to medical data and S/MIME for emails (OK, you could use PGPmail and some kind PGP Java authentication system, but that's quite unwieldy). While a small medical clinic may be able to setup a Certificate Authority (CA), you may need to use Verisign or another large CA to issue these certificates. But you're biggest stumbling block is user support. Installing browser certificates and educating the user on the safe use of certificates is quite a challenge.
I sure hope these new phones don't interfere with my Eternal Life Ring. I think the cold-fusion battery in the phone my just emit enough bogons...
I tend to like high tech solutions like verisign, but if I have to pay for a key, I'll look more closely at the free solutions first.
Try SwissSign for free X.509 certs and server certs.
When I heard that Sony was shipping 1/2 as many PS2s as planned for X-Mas I decided that I was going to have to buy a DVD player and stop watching movies on the computer. I did try one of the X10 DVD-Anywhere kits which was interesting, but...
So I started looking at players planning to spend about $150 and noticed that you can't get DTS at that price point. I started browsing the auctions and found an Onkyo DR90 at UBID.com for $349. It actually is a DTS Receiver/Amp with the DVD player built-in. I wanted a set of 5.1 speakers to go with it and UBID was selling the Warfedale Movie Modus 5 system for about $149. I'd never heard of Warfedale before but I figured the price was right. I was quite pleased with what I got. The Onkyo DR90 automatically selects the best settings for the DVD you are playing and the Warfedales are at least competent. The system may not be earth-shattering but for less than $500 I've been quite happy.
NOTE: The one feature this system lacks that will cause me to replace it all in a year or two is progressive scan. I never noticed how bad the interlacing problems were on DVD when I was playing movies on my computer!
Gee, I have a single config tool: it's called vi (on a minimal system).
Microsoft really f*cked up windows for many of us when they moved away from *.ini to the registry. Personally, I think flat-files for configuration are the pinnacle of configuration methods. XML might seem equally easy to edit, but the parsers would need to preserve formatting and comments. Why not just standardize on
# this is a comment
keyword=setting
We're almost there already and there are plenty of libraries out there to read and write them.
I guess the only real advantage of XML conf files would be that your GUI config utility wouldn't have to know about a new tool before it could give you the config options for it. But my philosophy is that if you have to use a GUI to config your system, maybe you shouldn't be config'ing it.
A few years back I was writing software to control tufting machines (like big sewing machines for making carpet). I almost used the GameBoy as the platform because it took almost zero processing power (the GameBoy has an 8Mhz Z80 clone - which is quite reasonable) and needed only about 64K to run, including the code and buffers to display a graphic of the patter. The GameBoy can actually address about 16MB of memory via page-flipping.
The real benefit of the GameBoy is if you try to price an industrial control with a few buttons and a nice LCD, you quickly exceed the $50 in small quantities that a GameBoy at Toys'R'Us costs.
Our HP NetServer LH6000's are running like a top. They did come with the wrong RAID BIOS, so Red Hat wouldn't recognize the RAID drives. Tech Support was very efficient and pointed us right to the problem. These are $40K+ boxes (6-way Xeon, 1GB RAM, 6x18GB RAID5). We are running Sybase 11.0.3 (about to move to 11.9.2) on them.
HP seems committed to providing the same level of support for Linux on Intel as they have for HPUX on PA-RISC. They are even offering Linux-oriented training courses through their educational programs.
Of course, supply on some parts, like the Xeons, is variable and the delivery time for one of these servers is way slow. And don't even think of running Mandrake (or really any distribution other than Red Hat) on the big boxes. The hardware support just isn't there.
We also have a couple HP LPr's, which have been selling on Onsale for cheap.
For smaller stuff, we have a slew of Penguin Computing 1RU servers. We order whatever Penguin has in stock because the delivery time is everything.
We also are using the Penguin Computing dual-head workstations for development systems. I'm not thrilled with how the Matrox G400 stretches a single desktop across both monitors (my dialogs pop up in the crack in between) but who am I to complain. Penguin also does a very good with their preinstalled distribution on their workstations. Upgraded kernel, Xi Graphics X Server, etc., etc.
Actually I would like to see a Samba port to NT. Samba provides much better control over Browse Master and PDC election than Windows SMB Server and it would be cool to be able to configure the thing with a text file.
;)
Of course, I don't want to see it bad enough to do the code myself.
Hmmm... My IBM 101Key keyboard I'm clacking away on right now is dated 11-NOV-91 (on the bottom). It's one of the newer one's I own. I started using these keyboards exclusively about five years ago. I used to try to just use whatever I was given, but once I started with the IBM keyboard I saw my accuracy and productivity go up. IBM has been making keyboards since before there were computers. The also made some of the first keyboards that didn't drive mechanical systems (i.e., electric typewriters). They really know what there are doing...
As far as keys in the penalty zones, I use CRiSP and find that my hands work in two modes: navigation/editting and creation. I guess this is analogous to edit and insert modes in vi. In navigation/edit mode my left hand hovers over the left shift key and my right hand hovers over the arrow keys. In edit mode my hands land on the home row.
CRiSP (like pretty much all editors I've used) provide ways to use the arrow keys to move more than one character. Ever try holding down the CTRL
key while left or right arrowing? It even works in
web browsers! And Home and End work pretty good too.
I really think the world is divided into two types: vi users and non-vi users. Personally I think vi is a single generation away from edlin and it's ilk. Give me windows, cut & paste, and modelessness, or give me death...
BTW, I use CRiSP in text mode more than I do in X windows. The beauty of CRiSP is all of the keystroke commands work identically in either CUI or GUI.
There are many languages that aren't inherently based on English, and more importantly, approach programming for very different points of view. Ever try forth? C++ provides operator and function overloading that could be used to create a non-English sublanguage.
But that's not the problem. Computers are procedural devices. Or rather, computers based on current CPU design are procedural. The processor executes one instruction, and then the next, and then the next. There are certain operations that this model requires: loops, branches, etc. Whether you call them "while", "for", "djnz", or something else, you still have the same basic concept. Even SMP and Massively Parallel systems are procedural - they just do more than one thing at once.
This is not an English thing. Changing the names of the commands won't change the way we write programs.
If you want to play around with a truly different programming paradigm, try neural networks.
And another note: I've worked with non-English speaking programmers. They tend to misspell variable names, but do it consistently!
Hmmm. I think are work (McKee Foods) I'd have problems with those names. First of all, we are forbidden to even mention Twinkies or Ho-hos. Second, if we named a machine Swissroll or NuttyBuddy or StarCrunch we'd confuse it for one of the production lines...
Instead we have really lame-o names like ORAD1 (Oracle Development 1), MFG1 (Manufacturing 1), etc...
Umm... Actually, the Air Force won't let you pilot a plane unless you have perfect 20/20 or better in both eyes uncorrected. I went through the academy physical in high school and they had to dilate my pupils for the vision test. I'm slightly off 20/20 in my left eye (couldn't even be labeled as 25/20, just NOT 20/20). It was enough to keep me from ever being able to fly.
Not only that, but you have to retest your senior year in the academy and still have 20/20 or better. Can you imagine the strain of four years of CompSci at the Air Force academy?
The Army has similar restrictions for helicopter pilots as does the Navy for Hydrofoil captains.
There are also many other restrictions for flying: height, weight (of course), you can never have blacked out in your life, and on and on...
Just kidding. I've been doing the Oracle thing for almost a year and am still astounded at just how bad SQL*Plus is. Unfortunately I use it everday. Most of my development is in C using Pro*C, so GUI tools aren't much help.
I am about to setup my own web server with an Oracle backend. I want to play around with WebDB and PHP3. I also need to play around with Developer because Oracle doesn't offer a certification program for 3GL programmers and I'm not ready to get into Oracle's Java mess.
BTW, PL/SQL, the Oracle procedural language, actually beat out Perl in my book as the most obscure language. The syntax is really funky and the use of semicolons is inconsistent with every language I've ever used.
But enough griping. If Oracle didn't make such screwy interfaces, people like me wouldn't make as much money making them work...
Remember that old GEOS operating system from the C64/C128? It grew into GEOS on PC in the late 80s, early 90s (the original AOL software used it!). It ran great on a 8088 with 640K RAM. It was an easy-to-use GUI that was legendary in it's robustness.
Guess what, it still exists:
http://www.newdealinc.com
It's grown up a little bit - best run on a 386 with at least 1MB of RAM, but still works on 286s. It includes a complete office suite and a web browser. There's even a BASIC interpreter.
The neatest feature it has is in the control panel you can set the user's skill level on a 1-4 range. All New Deal apps follow this setting.
A friend of mine who runs an ISP has New Deal on a 486 that his adopted daughter with Downs Syndrome uses. She's able to draw in the paint program, play solitaire (actually, she just moves the cards around), types in the word processor and browses the web.
It really doesn't get much easier. And as far as administration goes - anything that would kill New Deal would also take down Linux.
Gee, I wish you hadn't posted that as an AC...
And I think I should try to find another forum for this discussion. This is a VERY important topic for many Oracle folks...
My big question about this post is:
When you say it blows chunks, are you saying SQL*Plus blows chunks or does the server instance actually give up the ghost? I'll have to try it here at work (I think we've got an 8i instance around here someplace...)
Not quite. You can download RPMs of ASE 11.0.3 for development and evaluation purposes, but when you actually implement an application with it the license fees are the same as the NT version - about $300/user or a flat $50K if the app touches the web.
Oracle's fees are similar but they are pushing use of "application service providers". Basically web hosts that have licensed Oracle instances that you can use for a premium.
BTW, if you want a free set of Oracle Development CDs, check out:
http://www.oracleanswers.com
Register for the Atlanta or Birmingham seminars. They are fully booked - you won't get to go. But Oracle will send you a complimentary set of CDs for free
I've used PostgreSQL and Sybase for a few Linux based projects. I use Oracle 7.3 on HPUX at work. I'm helping a friend spec out a new project. He's ultimately going to use Oracle on Solaris (its government work - so the DB & OS choice is decided through politics). But development will likely occur on Oracle on Linux.
We have been going around in circles trying to decide whether to use Oracle 8.0.5 or 8i. As far as I can tell, the only significant advantage to 8i is the JVM. But 8i requires a minimum of 128MB RAM and 256MB swap without the JVM and 256MB RAM and 512MB with the JVM. Note: those are the minimums. And Oracle's not kidding. Don't even try firing up an 8i instance with less than 128MB RAM. Ultimately, the memory requirements will be satisfied. I suspect we'll spec out a Sun Ultra20 with about 1GB of RAM. But We don't want to invest a fortune in the development hardware.
Oracle 8.0.5, on the other hand, runs nicely with at little as 32MB of RAM. Plus, the install is CUI - whereas 8i requires X (I usually don't install X on my servers).
Besides a few cool new features in PL/SQL, does anyone else have any experience with 8.0.5 vs. 8i that would help sway me one way or the other? Right now we are pretty much set on using 8.0.5...
No RAW IO on Linux. I won't go into detail here, but Linux doesn't support Raw IO right now. Linus himself has stated that he has no interest in adding raw IO support because the performance increase from using buffered IO far exceeds the disaster recovery benefits of raw IO. But I hadn't though about FS limits. That's not so much a problem with Linux or Intel, but rather a problem with the POSIX file manipulation functions in GLIBC or LIBC. What does ftell return? A signed 32-bit long, which has a range of -2GB to +2GB. I think there are some kludges that use an unsigned long, but until 64 bit POSIX functions, we're stuck.
Anybody know if the 24 CPU box was going to support Linux?
IBM would have to throw some kernel hackers at the RS/6000 kernel in order to get it to tick. But nowadays that wouldn't be unheard of.
Reminds me alot of B.O.B. from the old Disney flick The Black Hole. If it runs WinCE, Microsoft could market it as: Microsoft B.O.B.!!!
To a certain degree, Microsoft does provide developer contact. They just don't publish it in "mainstream" press.
The Microsoft Developer's Network is the first line of contact with the developers: "here's a huge stack of CDs of everything we think you might need to develop for our OS."
You can also get significant interaction with developers and evangelists at trade shows. When I was programming games I found that the time spent at Microsoft Meltdown was invaluable. You get to schedule time in a hotel suite with DirectX developers from Microsoft to review your code and get pointers on how the OS works. Plus, it's their front-line for input on new features to add.
I was even invited to spent a few days in Redmond at the offices to get direct interaction with DirectX developers.
However, Microsoft is a traditional company when it comes to trade secrets. You didn't get invited to Meltdown without signing NDAs. OpenSource eliminates the need for NDAs and lawyers in general.
Microsoft's position is also bound by the fact that they are in the business of creating de facto standards. Linux is based on open standards. The concept of open community interaction is there from the initial stages of development.
My friend (who is reading over my shoulder) and I have been through two different game companies (actually three if you consider the first company before the buyout). While I've never enjoyed a job more, I don't think I'd want to get back into the market. The game market has really closed up a great deal compared to 5-10 years ago.
It's almost impossible for a new shop to take off. You pretty much have to develop a game completely before the distributors take you seriously and the game has to compare to what established shops were producing.
For instance, the first company I worked for made statistically accurate baseball and football games exclusively for the PC. Our boss wanted to go head-to-head with EA and try to add arcade style play and graphics. Well, the investment to make that leap is huge. We tried to do it with a team of 15 people - about 5 programmers and the rest artists and sports experts. We tried to turn out three titles a year - baseball, pro football and college football. We didn't even come close. I later interviewed with EA's division that does the Madden NFL games and found out that they have over 20 people working just on the pro football game alone.
Not to mention the disheartening sight of a streatch Hummer at E3 custom made for EA sports. That damned truck cost more than the production budget of our last title.
I too have moved on to ecommerce development and have experienced all of the benefits that are listed above. Better pay, reasonable hours, female coworkers, and on, and on.
I have also been asked dozens of times what it would take for someone to become a game developer. I always tell them - don't. It just isn't the career it used to be.
It is just the render engines. Pixar has a port of Renderman for Linux too (and they are working on the Alpha Linux port as we speak). And, of course, the Blue Moon Render Tools have existed on Linux for years. BMRT implements a subset of Renderman - I've messed around with exporting the geometries out of Alias on Irix and rendering on BMRT on Linux but didn't have enough management support to get beyond the initial test.
Alias released the renderer for NT about 3-4 months before the full Maya toolset. It was worth the wait. Of course, I'm more interested in seeing the good bits of Irix move to Linux. The release of SGI Linux 1.0 was more significant to me. If SGI put 1/10th the effort into Linux that they did Irix, NT would die quickly.
Ummm... If I remember right, Neural Nets are essential parallel-style nodes with extremely limited intelligence. It is the density of the interconnections between the nodes that actually create the "intelligence". Multiprocessor parallel systems are usually based on very intelligent nodes with limited interconnections (i.e., each node has a big Pentium or Alpha with local memory) and only one network connection.
Parallel Processing is like a think-tank. Lots of very bright people working on individual problems. Communication and coordination of the individuals becomes a major task and any problem submitted has to be diced up into individually bite-sized morsels.
Neural Nets is like a brain. A bunch of independently dumb cells with lots of interconnections. Communcation between the cells occurs "naturally" and the problem flows through the cells like an amorphous blob.
There is no easy way to take advantage of parallel processing. It has to be designed for early in the development process. What this library proposes is to write all applications for distributed processing, even if the software is normally run on single CPU or SMP boxes.
Of course, there are plenty of problems that are inherently easy to implement in parallel, brute-force crypto-cracking, scanning sections of data for coherent signals, packetized compression, graphics rendering, etc. Heck, even make can be parallelized easily.