Live and learn is right. I thought to disable the old sound, video, AGP and ACPI stuff but the IDE/ATA thing came at me from left field--never thought of that one! *sigh* it just seems so much easier to fix that kind of screw-up in Linux, Win9x and DOS
I have to disagree. I expect motherboard swaps to cause SOME turmoil. I DON'T expect them to completely crash and burn. Win NT/2K is the ONLY OS I've seen that couldn't at least boot into a single-user or safe mode and instead crashes completely.
As for being foolish--yes swapping motherboards would be if it was in a business environment. When it is your home PC I call it puttering around. I've puttered around like this with various versions of DOS, Win 3.1/95/98, Slakware, Red Hat, Mandrake, SCO... NOT ONE OF THESE had such a hard time with the brain transplant.
As for the merit of one OS installation to handle motherboard changes being "tangential to its merits as a usable system" consider a real situation I encountered: Some years ago at a student job, we had to take many PCs and upgrade from DOS/win3.1 to Windows 95 (back when 95 was still young and IE was just moved from the Plus Pack to be included with the OS). We did an install, set up all the standard apps and so on--on a single PC. We then made an image of that installation and duplicated it dozens of times, saving a huge amount of work (while honouring licensing terms of course:-). Not all of the PCs had identical motherboards. Some were brand new Pentium 90s, others were older 486 DX266s. They all had ATI graphics cards and 3COM NICs, but some were ISA and others were PCI, etc.
Of course very few of them ran perfectly right after the copy. However, the image restore was lightning fast compared even to an automated installation process, and more than made up for having to change device drivers (with the newer PCs it was grest because you just remove the bad drivers, reboot and PNP did unexpectedly well at picking up the correct hardware)--installing a few drivers from the setup CD was much easier than a full installation.
With Win2k, every time there was a different motherboard this method would result in the BSOD situation (some machines with IDENTICAL model numbers came factory assembled with different motherboard revisions or IDE controllers, etc). Given this useful method of deployment and my propensity to "foolishly" swap motherboards annually on my home machines, I indeed DO consider the ability to handle hardware changes an important merit.
Of course, I'm not a professional IT guy. I do estimates and budgets. Sometimes I program. Sometimes I build digital circuits. I haven't been an IT guy since my student days. Perhaps Microsoft has improved deployment of Windows via networking in the past 5 years (of course if our servers aren't windows you'd still be SOL I bet). I don't deal with the IT guys on a day to day basis and so haven't had the opportunity to see if it works as well as the drive image thing does...
OK, I'll kick this dead horse one more time. I 'dug up' the floppy and made a repair disk and put it away for safe keeping. It was not close at hand because I wasn't planning on using it the next day. It was an upgrade of a home PC I was planning to make after new years, until I came across a really good Boxing Day special that evening. I still think going into a closet in another room to get the W2K CD and emergency disk classifies as "digging up"--especially if you see how my closet is arranged.
As for the BSOD--it wasn't before getting to safe mode. I selected Safe Mode--THEN--seconds after going to graphics mode--BSOD (safe mode worked before the upgrade too). I had to boot from CD into the Recovery Console before I could do anything at all. Honestly, I'm not full of crap--I'm just not a Microsoft guru. I'm more at home in the hardware end of things and did the bulk of my school and work on UNIX-style systems. For the record I curse my Linux box (where I passed the old W2K motherboard to--was loathe to disassemble a newly upgraded runnung Linux box to boot my W2K again). Except with Mandarke 7.2, I always had a bitch of a time getting my 3DFX Banchee going. With Linux though I get over it faster because it's free--I just sigh and say "what can you expect for free" and dig in.
If you are indeed a MS evangelist, please don't be offended. I have to say that once all the crap and rebooting is done, everything works great, and MS has done volumes to increase reliability and ease of use in comparison to previous versions. I'm just very critical about products I've spend a lot of money on.
As instructed in the slim Win2K manual, I removed drivers for hardware specific to my motherboard (sound, AGP, etc) but missed the IDE controller stuff. I even booted off of CD into an emergency command-prompt mode and tried picking around in there. Still BSODed. In desperation I did the repair/re-install.
Also, I never said I was surprised at it overwriting all my updates. I fully expected it. My beef is that I was reduced to doing the restore for new hardware in the first place. I expected more from a highly touted, expensive operating system (OK, not as expensive as some commercial UNIX solutions, but pricy for me nonetheless). At least it could have let me in in "safe mode" (it BSODed even then) using generic drivers (if Linux can have them why not Windows?). I was further aggrivated by the need to reboot after EACH HOTFIX. I expect the odd one if they are kernel related, but to patch a security hole in the browser?
As for the twenty reboods you are almost right--it was SP1, upgrade to strong encryption, IE 5.0 to 5.5 and 14 hotfixes.
There is absolutely NO excuse for requiring a re-install of the operationg system when hardware changes. In my opinion, that is a serious liability. What happens if you have a motherboard failure (stuff like that happens), and the motherboard is discontinued, or now ships with another bios or is a newer revision. I agree totally that motherboards are not all the same (I'm not that naive--I've designed microprocessor-based systems and written BIOS routines, etc. I know that there can be significant, important differences in motherboards, even if they are the exact same models but different revisions--this is reality even in much simpler systems like I worked on). Also, What if you want to make several "Images" of a drive to ease installation of W2K Pro, but can't count on all the PC's to be IDENTICAL?
Obviously, for mission-critical stuff, you would have a backup server in the event of hardware failure, but you still need to restore the primary machine. Instead of puking out a STOP 0x7b when it encounters a certain driver/hardware mismatch (as has often happened since NT 3.51) why can't Win2k catch the exception and continue with the "generic" drivers, and allow the user to correct the problem without an invasive repair/re-install process? THERE IS NO GOOD REASON why it can't. Microsoft already allows you to boot in "VGA Mode" if you change video cards. This could be extended to all subsystems.
Yes, there ARE countless different motherboards. However, all decent ones have the same entry points into the black box. They all have PCI and AGP slots, they all support the same BIOS calls and so on. Why can't Win2K (even if only in "Safe Mode") deal with the common denominator in the event the "optimised" configuration no longer applies. I know it seems harsh to hold Win2K to those standards. However, if I'm a company paying tens of thousands a year in licensing fees, a "smarter boot sequence" to save many hours of IT work isn't just a "nice" thing. If the FREE competition appears to be superior in that respect it would be something I'd DEMAND in return for those thousands I gave to Microsoft--particularly then they won't even give me the tools (ie. source code) to make improvements myself (improvements I'd be willing to share).
Yes, I do have a double standard w.r.t. Win2k vs. Linux. It's because they are two different situations. Linux isn't "perfect" or "optimised for everything". However, it is free (monetarily) so you get more than you pay for. It is also Free (open source) so I can contribute improvements. Until MS Windows is Free (open source if not monitarily) I will not be nearly as forgiving with it's shortcomings.
</rant>
It's nice to see a comprehensive critique of a Linux distribution--it doesn't happen often enough. The biggest complaint about Linux (or excuse used to defend Microsoft Windows) I hear from everyday users is that Linux has so many distributions, and they are all different and difficult to install and use.
Mandrake has to be my favourite distribution to date (I've also tried Red Hat and Slakware). LM 7.2 was very easy to install and every version of LM has shown improvement. Although Slakware or Debian might be the best for stability, LM has been good in this area while focusing on ease-of-use.
I just went through the exercise of upgrading hardware (MB/CPU only) of Win2K and LM 7.2 machines. Upon upgrading the LM 7.2 machine, it figured out the new hardware config. automatically. Win2k was a different matter. It didn't like the new on-board IDE controller and blue-screened upon bootup. I had to dig out the recovery disk, boot from the install CD and "Repair" my installation to bring it back. The "Repair" undid Service Pack 1 and all the hotfixes. TWENTY reboots later (that is not an exaggeration!) I was back in the saddle. Anyone else had the same experience upgrading a motherboard in a Win2K machine? So much for Windows ease-of-use over Linux...
A couple of notes about the review: It claimed there was not an Apache package in the 7.2 distribution. This is not true. I obtained my LM 7.2 by downloading 2 ISO images--the setup/install CD and the "extensions" CD. Much of the good stuff (notably Apache and PostgreSQL) were on the "extensions" CD. During setup you are asked if you have the extensions CD and are prompted th enter it at the appropriate time. I strongly recommend using the Apache packages for 7.2 over 7.1 since they are updated (some bugs and file permission settings fixed) and tuned to conserve memory when using mod_perl. Also, if you choose "high" or "paranoid" security level take heed of the warning Mandrake gives you--these are best used for servers because many of the security measures break some of Mandrake's GUI-based eye-candy and cause slightly different behaviour from documentation.
Can't wait for the review of Debian (that's the one I want to try next)...
I feel compelled to correct our Commie fan. The 800XL did NOT max out at 48K, it came STANDARD with a full 64K or RAM. Remember, the 6502/6510 CPU could only address 64K at once and both the 800XL and C64 used the same family of processors. This means that additional ROM had to be mapped in place of part of the RAM when the machines booted up in BASIC.
In the 800XL 48K of memory was free for programs when all the ROM was switched in place (I'll concede that if DOS was loaded in with it that figure shrank slightly). On the C64, LESS THAN 40K was free. To top it all off, all the 8-bit Ataris ran at an 80% HIGHER clock speed than the C64, and the 800XL had 256 colours vs. the C64 (the Atari could produce some stunning greyscale images). The Atari disc drives were something like 10 times faster. In fact, the C64 drive was considerably SLOWER than the CASSETTE TAPE drive on another interesting 8-bit computer, the Coleco ADAM (I personally tested this fact and found it to be true). I/O in general was better in every way on the Atari XL series (faster, easier to program, more expandible) than on the C64.
Atari BASIC actually had sound and graphics commands (the C64, desipte it's sound and graphics power had a sucky Commodore PET-derived BASIC). Atari disc drive users also had a REAL DOS. On the C64 you had to use arcane, kludgy commands to even get a simple directory listing.
And last but not least, C64s and the earlier 1541 disc drives were less reliable than anything Atari ever put out. The acid test is how they fared in schools. The C64's (unlike the rock-solid PETs of an earlier era) were always suffering from joystick connector problems, power supply glitches, keyboard defects and do on. Ataris seemed to fare better under similar use (even though the keyboards had a mushy feel that would frustrate touch typists).
In conclusion, Commies suck and Atari Rulez d00dz!
Really, XML technology is so new I don't believe either or any XML standard or framework is established yet. In my experience with manufacturing and warehousing operations I've found that a surprisingly large number of them are still stuck on some expensive, proprietary VAX or IBM mainframe application that talks to nothing else, or even stick to pure paper systems. I've seen office assistants print off reports from the VAX and HAND TYPE the results into Excel to do sorts, calculations, graphs and so on--ridiculous!
What some people seem to lose sight of is that the whole concept of XML is extremely new if you can manage to see it from the perspective of everyone except geeks. The market for ebXML or BizTalk is almost completely untapped and too immature to make pronouncements on what will be in the future. Even Microsoft and IBM know that--while they are the prime backers of BizTalk, take a look under "I" and "M" in the ebXML List of Participants. That's right, there are contingents from both of them. How's that for hedging your bets?
I have to admit that I was concerned about my decision to go with DSL, but my experiences have been extremely positive. Perhaps it is how it is being implemented in the US and that in most of the US it is a relatively new service. (Hmm... I should be careful here--I don't mean to dump on the US. Fellow Canadians who dump on the US is a pet peeve of mine. Truthfully, DSL is a fairly new and untested service in many smaller Canadian cities as well. It's just that Calgary, Ottawa, Toronto and Vancouver are very wired cities even on a global scale, just as most of California is in the US)
I live in Calgary, where DSL service has been available publicly since 1997 (and for a short time before that in limited downtown locations). I myself have had the service for 9 months so far and have only suffered about 12 hours downtime in 2 outages during the entire time.
At the time I was looking at broadband access, Shaw cable was severely over-subscribing their service in some areas of the city and had recent nameserver outages. Add to the fact that if you mentioned Linux they would go into yellow alert--warning you that they could offer no support and that if you were caught running servers you could be cut off. They would still be glad to hook you up, but don't expect them to help you if something's not working. When it runs at it's best it would download at about 2Mb/s but when the pipes clogged it would go down to 50-100Kb/s--then you would lose service entirely for a day when they upgraded and you got full speed again. This was happening about every 4 weeks to a co-worker. Perhaps those "net hog" commercials aren't entirely true but they aren't complete lies either. Basically if you wanted real support, fixed IP's for "officially approved" servers and consistent service from Shaw (good enough for business or serious hobby activity, you had to opt for Shaw FibreLink which is quite a bit more expensive.
Contrast this with my DSL experience. For about the equivalent of US$20-25 per month you get "4000K/500K" service, or for about $65-70 per month "7000K/1000K basic business" service with a fixed 8 IP subnet (1 network, 1 broadcast, 1 modem and 5 free IPs) through PSI/CADvision. Although PSI/CADVision seems to advertise the highest maximum speeds, I've heard that pretty much all the DSL services (Telus, OANet, etc) in reality are about the same speed and reliability--about 1-2 Mb/s for downloads and 300-500 Kb/s for uploads.
As for service I can't complain. They promised connection in 4 to 6 weeks and were done in 5. PSI/CADvision doesn't officially support Linux, and when the installer came to set up my modem he was clueless about Linux. However, he easily got the connection going on a Windows box, and supplied me with all the right information in hard copy to set up my Linux box. He was curious about Linux though and stuck around to watch me set it up (I got the basic business plan, so it was trivial since I had a static IP). Their tech support people are quite knowledgeable about networking in general, and their web-based "toolkits" are great (will test to see if you've correctly restricted mail relay, has a web interface to NMap to scan your hosts for open ports, do all your forward and reverse DNS on their nameservers, e-mail account config and everything...)
I know this comes off as a salesman pitching a product, but I stress that I stand to gain nothing by promoting PSI/CADVision. From what I've heard ALL the main DSL service providers in Calgary offer about the same in performance and reliability, and I assume the same would be true in other cities--speed/reliability is nearly the same within a specific region (bad or good) regardless of who offers it.
Through work I've also been able to see what Telus offers. Performance/reliability wise they were comparable, and their on-site installation and service people were great--they were competent and worked hard at resolving our problems, even if they weren't the fastest for service (The install waiting time was actully slightly quicker than CADVision's though). Their on-line, e-mail and phone support was another matter--very slow response times and you were passed around to many people. They were friendly enough but didn't seem to know which of their many departments was supposed to handle my requests. Also, their on-line tools were not nearly as complete or as well implements as PSI/CADVisions, which placed a heavier demand on support staff. Fortunately once the service was set up and running you didn't need to deal with them much. Overall they were quite acceptable as well
Basically, I would look at two factors in choosing a DSL provider. First, consider the ones who have offered the service the longest. They are likely to have the most reliable service as they have worked out all the kinks. Second, look for the bigger players--the small ones often stretch their resources too thin. Telco's (which is what Telus is) would brobably have the most reliable service and reasonable speed (and in Telus' case at least had a relatively short instalation waiting period), but expect typical telco customer service.
I hope this provides some insight to those looking for DSL service, even if it is very western-Canadian oriented. At least it set's the record straight in that there is such a thing as fast, dependable DSL service...
Given the power and availability of off-the-shelf PC hardware technology, I'm surprised that PC assemblers haven't thought of such an idea. You don't need much in the way of proprietary technology to make the ultimate video recorder. Here is my modest proposal:
First of all, to get a compact form factor use a case and motherboard that employ the NLX form factor. Unlike ATX, this standard employs a 3-slot PCI riser card to allow a slimline design. It is also very serviceable--the motherboard could be removed and replaced without removing any cards, and many cases are built for such home-entertainment features as front infra-red ports and designs that compliment stereo systems. There are some very nice NLX cases out there that are the same size as most VCR's and would fit in your TV cabinet nicely.
Use video display/capture cards with TV in/out like ATI's all-in-wonder, and hardware DVD/MPEG decoders like Netstream 2000. Hardware DVD and MPEG decoders reduce the CPU requirements and have the added benefit of being "MPAA-approved" if you care about such things. This is a video recorder, so you don't need a screaming high-end machine or hugh 3D acceleration--money saved there can go towards a massive 40GB drive if you wish.
At this point you can go beyond TiVO:
Add a DVD player and you can watch rented videos. Add a CD-RW, IOmega JAZZ or Castlewood ORB drive and you have removable media so you can tape the Olympic opening ceremonies and keep them and record over the copy on your hard drive. Alternatively, add FireWire and you could digitally transfer video to and from your miniDV camcorder
As for software: if Linux is good enough for TiVO it's good enough for me. Either GNOME or KDE-based environments could be used, with a window manager optimised for TV screens (the GNOME desktops look particularly nice). Someone could start an Open Source project at develop a standard XML application for TV Listings. The TV Listings could be made available via freenet or GNUtella. The advantage to such an open system, besides being free in all senses of the word, is that it doesn't rely on a corporation being around to provide the service. You risk your privacy since they can monitor your viewing habits, and if they lost out to a competitor and go out of business you could end up with the next Betamax VCR--unless someone deems their market share is large enough to take over the service or is generous enough to open the system up.
So get out there and build the ultimate video appliance!
Slashdot Mirror
Live and learn is right. I thought to disable the old sound, video, AGP and ACPI stuff but the IDE/ATA thing came at me from left field--never thought of that one! *sigh* it just seems so much easier to fix that kind of screw-up in Linux, Win9x and DOS
I have to disagree. I expect motherboard swaps to cause SOME turmoil. I DON'T expect them to completely crash and burn. Win NT/2K is the ONLY OS I've seen that couldn't at least boot into a single-user or safe mode and instead crashes completely.
:-). Not all of the PCs had identical motherboards. Some were brand new Pentium 90s, others were older 486 DX266s. They all had ATI graphics cards and 3COM NICs, but some were ISA and others were PCI, etc.
As for being foolish--yes swapping motherboards would be if it was in a business environment. When it is your home PC I call it puttering around. I've puttered around like this with various versions of DOS, Win 3.1/95/98, Slakware, Red Hat, Mandrake, SCO... NOT ONE OF THESE had such a hard time with the brain transplant.
As for the merit of one OS installation to handle motherboard changes being "tangential to its merits as a usable system" consider a real situation I encountered: Some years ago at a student job, we had to take many PCs and upgrade from DOS/win3.1 to Windows 95 (back when 95 was still young and IE was just moved from the Plus Pack to be included with the OS). We did an install, set up all the standard apps and so on--on a single PC. We then made an image of that installation and duplicated it dozens of times, saving a huge amount of work (while honouring licensing terms of course
Of course very few of them ran perfectly right after the copy. However, the image restore was lightning fast compared even to an automated installation process, and more than made up for having to change device drivers (with the newer PCs it was grest because you just remove the bad drivers, reboot and PNP did unexpectedly well at picking up the correct hardware)--installing a few drivers from the setup CD was much easier than a full installation.
With Win2k, every time there was a different motherboard this method would result in the BSOD situation (some machines with IDENTICAL model numbers came factory assembled with different motherboard revisions or IDE controllers, etc). Given this useful method of deployment and my propensity to "foolishly" swap motherboards annually on my home machines, I indeed DO consider the ability to handle hardware changes an important merit.
Of course, I'm not a professional IT guy. I do estimates and budgets. Sometimes I program. Sometimes I build digital circuits. I haven't been an IT guy since my student days. Perhaps Microsoft has improved deployment of Windows via networking in the past 5 years (of course if our servers aren't windows you'd still be SOL I bet). I don't deal with the IT guys on a day to day basis and so haven't had the opportunity to see if it works as well as the drive image thing does...
OK, I'll kick this dead horse one more time. I 'dug up' the floppy and made a repair disk and put it away for safe keeping. It was not close at hand because I wasn't planning on using it the next day. It was an upgrade of a home PC I was planning to make after new years, until I came across a really good Boxing Day special that evening. I still think going into a closet in another room to get the W2K CD and emergency disk classifies as "digging up"--especially if you see how my closet is arranged.
As for the BSOD--it wasn't before getting to safe mode. I selected Safe Mode--THEN--seconds after going to graphics mode--BSOD (safe mode worked before the upgrade too). I had to boot from CD into the Recovery Console before I could do anything at all. Honestly, I'm not full of crap--I'm just not a Microsoft guru. I'm more at home in the hardware end of things and did the bulk of my school and work on UNIX-style systems. For the record I curse my Linux box (where I passed the old W2K motherboard to--was loathe to disassemble a newly upgraded runnung Linux box to boot my W2K again). Except with Mandarke 7.2, I always had a bitch of a time getting my 3DFX Banchee going. With Linux though I get over it faster because it's free--I just sigh and say "what can you expect for free" and dig in.
If you are indeed a MS evangelist, please don't be offended. I have to say that once all the crap and rebooting is done, everything works great, and MS has done volumes to increase reliability and ease of use in comparison to previous versions. I'm just very critical about products I've spend a lot of money on.
The repair disk was made the day before I upgraded.
;-)
Maybe I would have disabled custom hardware in safe mode--if I could have gotten to safe mode. Safe mode BSODed on me too.
Microsoft Apologists, eh?
As instructed in the slim Win2K manual, I removed drivers for hardware specific to my motherboard (sound, AGP, etc) but missed the IDE controller stuff. I even booted off of CD into an emergency command-prompt mode and tried picking around in there. Still BSODed. In desperation I did the repair/re-install.
Also, I never said I was surprised at it overwriting all my updates. I fully expected it. My beef is that I was reduced to doing the restore for new hardware in the first place. I expected more from a highly touted, expensive operating system (OK, not as expensive as some commercial UNIX solutions, but pricy for me nonetheless). At least it could have let me in in "safe mode" (it BSODed even then) using generic drivers (if Linux can have them why not Windows?). I was further aggrivated by the need to reboot after EACH HOTFIX. I expect the odd one if they are kernel related, but to patch a security hole in the browser?
As for the twenty reboods you are almost right--it was SP1, upgrade to strong encryption, IE 5.0 to 5.5 and 14 hotfixes.
There is absolutely NO excuse for requiring a re-install of the operationg system when hardware changes. In my opinion, that is a serious liability. What happens if you have a motherboard failure (stuff like that happens), and the motherboard is discontinued, or now ships with another bios or is a newer revision. I agree totally that motherboards are not all the same (I'm not that naive--I've designed microprocessor-based systems and written BIOS routines, etc. I know that there can be significant, important differences in motherboards, even if they are the exact same models but different revisions--this is reality even in much simpler systems like I worked on). Also, What if you want to make several "Images" of a drive to ease installation of W2K Pro, but can't count on all the PC's to be IDENTICAL?
Obviously, for mission-critical stuff, you would have a backup server in the event of hardware failure, but you still need to restore the primary machine. Instead of puking out a STOP 0x7b when it encounters a certain driver/hardware mismatch (as has often happened since NT 3.51) why can't Win2k catch the exception and continue with the "generic" drivers, and allow the user to correct the problem without an invasive repair/re-install process? THERE IS NO GOOD REASON why it can't. Microsoft already allows you to boot in "VGA Mode" if you change video cards. This could be extended to all subsystems.
Yes, there ARE countless different motherboards. However, all decent ones have the same entry points into the black box. They all have PCI and AGP slots, they all support the same BIOS calls and so on. Why can't Win2K (even if only in "Safe Mode") deal with the common denominator in the event the "optimised" configuration no longer applies. I know it seems harsh to hold Win2K to those standards. However, if I'm a company paying tens of thousands a year in licensing fees, a "smarter boot sequence" to save many hours of IT work isn't just a "nice" thing. If the FREE competition appears to be superior in that respect it would be something I'd DEMAND in return for those thousands I gave to Microsoft--particularly then they won't even give me the tools (ie. source code) to make improvements myself (improvements I'd be willing to share).
Yes, I do have a double standard w.r.t. Win2k vs. Linux. It's because they are two different situations. Linux isn't "perfect" or "optimised for everything". However, it is free (monetarily) so you get more than you pay for. It is also Free (open source) so I can contribute improvements. Until MS Windows is Free (open source if not monitarily) I will not be nearly as forgiving with it's shortcomings.
</rant>
It's nice to see a comprehensive critique of a Linux distribution--it doesn't happen often enough. The biggest complaint about Linux (or excuse used to defend Microsoft Windows) I hear from everyday users is that Linux has so many distributions, and they are all different and difficult to install and use.
Mandrake has to be my favourite distribution to date (I've also tried Red Hat and Slakware). LM 7.2 was very easy to install and every version of LM has shown improvement. Although Slakware or Debian might be the best for stability, LM has been good in this area while focusing on ease-of-use.
I just went through the exercise of upgrading hardware (MB/CPU only) of Win2K and LM 7.2 machines. Upon upgrading the LM 7.2 machine, it figured out the new hardware config. automatically. Win2k was a different matter. It didn't like the new on-board IDE controller and blue-screened upon bootup. I had to dig out the recovery disk, boot from the install CD and "Repair" my installation to bring it back. The "Repair" undid Service Pack 1 and all the hotfixes. TWENTY reboots later (that is not an exaggeration!) I was back in the saddle. Anyone else had the same experience upgrading a motherboard in a Win2K machine? So much for Windows ease-of-use over Linux...
A couple of notes about the review: It claimed there was not an Apache package in the 7.2 distribution. This is not true. I obtained my LM 7.2 by downloading 2 ISO images--the setup/install CD and the "extensions" CD. Much of the good stuff (notably Apache and PostgreSQL) were on the "extensions" CD. During setup you are asked if you have the extensions CD and are prompted th enter it at the appropriate time. I strongly recommend using the Apache packages for 7.2 over 7.1 since they are updated (some bugs and file permission settings fixed) and tuned to conserve memory when using mod_perl. Also, if you choose "high" or "paranoid" security level take heed of the warning Mandrake gives you--these are best used for servers because many of the security measures break some of Mandrake's GUI-based eye-candy and cause slightly different behaviour from documentation.
Can't wait for the review of Debian (that's the one I want to try next)...
I feel compelled to correct our Commie fan. The 800XL did NOT max out at 48K, it came STANDARD with a full 64K or RAM. Remember, the 6502/6510 CPU could only address 64K at once and both the 800XL and C64 used the same family of processors. This means that additional ROM had to be mapped in place of part of the RAM when the machines booted up in BASIC.
In the 800XL 48K of memory was free for programs when all the ROM was switched in place (I'll concede that if DOS was loaded in with it that figure shrank slightly). On the C64, LESS THAN 40K was free. To top it all off, all the 8-bit Ataris ran at an 80% HIGHER clock speed than the C64, and the 800XL had 256 colours vs. the C64 (the Atari could produce some stunning greyscale images). The Atari disc drives were something like 10 times faster. In fact, the C64 drive was considerably SLOWER than the CASSETTE TAPE drive on another interesting 8-bit computer, the Coleco ADAM (I personally tested this fact and found it to be true). I/O in general was better in every way on the Atari XL series (faster, easier to program, more expandible) than on the C64.
Atari BASIC actually had sound and graphics commands (the C64, desipte it's sound and graphics power had a sucky Commodore PET-derived BASIC). Atari disc drive users also had a REAL DOS. On the C64 you had to use arcane, kludgy commands to even get a simple directory listing.
And last but not least, C64s and the earlier 1541 disc drives were less reliable than anything Atari ever put out. The acid test is how they fared in schools. The C64's (unlike the rock-solid PETs of an earlier era) were always suffering from joystick connector problems, power supply glitches, keyboard defects and do on. Ataris seemed to fare better under similar use (even though the keyboards had a mushy feel that would frustrate touch typists).
In conclusion, Commies suck and Atari Rulez d00dz!
Really, XML technology is so new I don't believe either or any XML standard or framework is established yet. In my experience with manufacturing and warehousing operations I've found that a surprisingly large number of them are still stuck on some expensive, proprietary VAX or IBM mainframe application that talks to nothing else, or even stick to pure paper systems. I've seen office assistants print off reports from the VAX and HAND TYPE the results into Excel to do sorts, calculations, graphs and so on--ridiculous!
What some people seem to lose sight of is that the whole concept of XML is extremely new if you can manage to see it from the perspective of everyone except geeks. The market for ebXML or BizTalk is almost completely untapped and too immature to make pronouncements on what will be in the future. Even Microsoft and IBM know that--while they are the prime backers of BizTalk, take a look under "I" and "M" in the ebXML List of Participants. That's right, there are contingents from both of them. How's that for hedging your bets?
I have to admit that I was concerned about my decision to go with DSL, but my experiences have been extremely positive. Perhaps it is how it is being implemented in the US and that in most of the US it is a relatively new service. (Hmm... I should be careful here--I don't mean to dump on the US. Fellow Canadians who dump on the US is a pet peeve of mine. Truthfully, DSL is a fairly new and untested service in many smaller Canadian cities as well. It's just that Calgary, Ottawa, Toronto and Vancouver are very wired cities even on a global scale, just as most of California is in the US)
I live in Calgary, where DSL service has been available publicly since 1997 (and for a short time before that in limited downtown locations). I myself have had the service for 9 months so far and have only suffered about 12 hours downtime in 2 outages during the entire time.
At the time I was looking at broadband access, Shaw cable was severely over-subscribing their service in some areas of the city and had recent nameserver outages. Add to the fact that if you mentioned Linux they would go into yellow alert--warning you that they could offer no support and that if you were caught running servers you could be cut off. They would still be glad to hook you up, but don't expect them to help you if something's not working. When it runs at it's best it would download at about 2Mb/s but when the pipes clogged it would go down to 50-100Kb/s--then you would lose service entirely for a day when they upgraded and you got full speed again. This was happening about every 4 weeks to a co-worker. Perhaps those "net hog" commercials aren't entirely true but they aren't complete lies either. Basically if you wanted real support, fixed IP's for "officially approved" servers and consistent service from Shaw (good enough for business or serious hobby activity, you had to opt for Shaw FibreLink which is quite a bit more expensive.
Contrast this with my DSL experience. For about the equivalent of US$20-25 per month you get "4000K/500K" service, or for about $65-70 per month "7000K/1000K basic business" service with a fixed 8 IP subnet (1 network, 1 broadcast, 1 modem and 5 free IPs) through PSI/CADvision. Although PSI/CADVision seems to advertise the highest maximum speeds, I've heard that pretty much all the DSL services (Telus, OANet, etc) in reality are about the same speed and reliability--about 1-2 Mb/s for downloads and 300-500 Kb/s for uploads.
As for service I can't complain. They promised connection in 4 to 6 weeks and were done in 5. PSI/CADvision doesn't officially support Linux, and when the installer came to set up my modem he was clueless about Linux. However, he easily got the connection going on a Windows box, and supplied me with all the right information in hard copy to set up my Linux box. He was curious about Linux though and stuck around to watch me set it up (I got the basic business plan, so it was trivial since I had a static IP). Their tech support people are quite knowledgeable about networking in general, and their web-based "toolkits" are great (will test to see if you've correctly restricted mail relay, has a web interface to NMap to scan your hosts for open ports, do all your forward and reverse DNS on their nameservers, e-mail account config and everything...)
I know this comes off as a salesman pitching a product, but I stress that I stand to gain nothing by promoting PSI/CADVision. From what I've heard ALL the main DSL service providers in Calgary offer about the same in performance and reliability, and I assume the same would be true in other cities--speed/reliability is nearly the same within a specific region (bad or good) regardless of who offers it.
Through work I've also been able to see what Telus offers. Performance/reliability wise they were comparable, and their on-site installation and service people were great--they were competent and worked hard at resolving our problems, even if they weren't the fastest for service (The install waiting time was actully slightly quicker than CADVision's though). Their on-line, e-mail and phone support was another matter--very slow response times and you were passed around to many people. They were friendly enough but didn't seem to know which of their many departments was supposed to handle my requests. Also, their on-line tools were not nearly as complete or as well implements as PSI/CADVisions, which placed a heavier demand on support staff. Fortunately once the service was set up and running you didn't need to deal with them much. Overall they were quite acceptable as well
Basically, I would look at two factors in choosing a DSL provider. First, consider the ones who have offered the service the longest. They are likely to have the most reliable service as they have worked out all the kinks. Second, look for the bigger players--the small ones often stretch their resources too thin. Telco's (which is what Telus is) would brobably have the most reliable service and reasonable speed (and in Telus' case at least had a relatively short instalation waiting period), but expect typical telco customer service.
I hope this provides some insight to those looking for DSL service, even if it is very western-Canadian oriented. At least it set's the record straight in that there is such a thing as fast, dependable DSL service...
Given the power and availability of off-the-shelf PC hardware technology, I'm surprised that PC assemblers haven't thought of such an idea. You don't need much in the way of proprietary technology to make the ultimate video recorder. Here is my modest proposal:
First of all, to get a compact form factor use a case and motherboard that employ the NLX form factor. Unlike ATX, this standard employs a 3-slot PCI riser card to allow a slimline design. It is also very serviceable--the motherboard could be removed and replaced without removing any cards, and many cases are built for such home-entertainment features as front infra-red ports and designs that compliment stereo systems. There are some very nice NLX cases out there that are the same size as most VCR's and would fit in your TV cabinet nicely.
Use video display/capture cards with TV in/out like ATI's all-in-wonder, and hardware DVD/MPEG decoders like Netstream 2000. Hardware DVD and MPEG decoders reduce the CPU requirements and have the added benefit of being "MPAA-approved" if you care about such things. This is a video recorder, so you don't need a screaming high-end machine or hugh 3D acceleration--money saved there can go towards a massive 40GB drive if you wish.
At this point you can go beyond TiVO: Add a DVD player and you can watch rented videos. Add a CD-RW, IOmega JAZZ or Castlewood ORB drive and you have removable media so you can tape the Olympic opening ceremonies and keep them and record over the copy on your hard drive. Alternatively, add FireWire and you could digitally transfer video to and from your miniDV camcorder
As for software: if Linux is good enough for TiVO it's good enough for me. Either GNOME or KDE-based environments could be used, with a window manager optimised for TV screens (the GNOME desktops look particularly nice). Someone could start an Open Source project at develop a standard XML application for TV Listings. The TV Listings could be made available via freenet or GNUtella. The advantage to such an open system, besides being free in all senses of the word, is that it doesn't rely on a corporation being around to provide the service. You risk your privacy since they can monitor your viewing habits, and if they lost out to a competitor and go out of business you could end up with the next Betamax VCR--unless someone deems their market share is large enough to take over the service or is generous enough to open the system up.
So get out there and build the ultimate video appliance!
--
No sig for you!!