On 7/20/10, Liam Proven <lproven at gmail.com> wrote: It's not just Windows - debugging the stack under Linux isn't any better. At least with libusb, I have the source and can inject debugging printf()s into it. With a serial port, you can turn an old machine with two serial interfaces into a snooper and capture the traffic, or use an old HP4951 protocol analyzer (I have more than one) which is _really_ designed to snoop sync, async, RS-232, RS-423, etc. We even used one to simulate an IBM PU Type 4 to trace our product's code (a PU Type 2 emulator) and how it handled various bits in the BIND sequence. Given the speeds available, that was not a typical use for serial ports, though I have done plenty of sync and async networking (X.25, DDCMP, HASP, 3780, SNA). Agreed. I find SCSI, Firewire and eSATA all superior for attaching external storage. We had to move away from USB to eSATA at work, partially due to a lack of USB support via enterprise-grade VMware (mentioned in another message) and partially due to speed. Copying 20 million files took *4 days* over USB and switching to eSATA on the same host reduced the copy time to 18 hours. I also mentioned earlier that I'm in the middle of a copy that started last night of 1TB that was going to take 55 hours via USB but dropped to 31 hours when someone with a Firewire 800 cable in her bag literally just happened to walk in at the right moment. Firewire is no harder to plug in or configure or use than USB, but as you pointed out earlier, is unfortunately dying out. My old laptop had real serial, USB 2.0 and Firewire. My new laptop has USB 2.0 and eSATA. I miss my old laptop (though I do think eSATA is one of the few actual improvements, vs feature curtailments, we've seen lately). Generic cables? The ends are generic, sure, but there's A, B, mini A, several types of mini B (one dominant, but hardly unique), and then there's USB 1.1 vs USB 2.0 in terms of cable ratings (but will a USB 1.1-marked cable pushed to 2.0 speeds really be too noisy or is that just marketing?) I have no less than 5 flavors of USB cables lying around the house, though two of the 5 are the most common. SCSI is its own nightmare, especially in the early days, and doubly so when vendors decide to get cute (like Apple's non-standard DB25 SCSI connector on the Mac Plus that turned into a standard for SCSI-1 but couldn't handle Fast or Wide, etc). I've hooked up a lot of SCSI drives and in the enterprise world, where people spend real money for products that better "just work", they do. The consumer arena is another matter entirely and there are lots of short cuts and wild variations from vendor to vendor that make the situation worse. On the desktop, when I was the one placing the orders, I bought Adaptec and sidestepped a lot of minor-player driver and termination and connector hassles. Unfortunately, not all of my associates did the same, so I did have messes to contend with. SCSI _is_ standardized, but not everyone decided to follow the standard. That actually brings up another annoyance I have with USB - it's trying to replace several interfaces at once - RS-232, PS/2 and SCSI - for connecting peripherals to your machine. Insofar as it's trying to cover all those bases, it does a passable job, but I think it merely fills the gap passably - with the exception of reducing the variety of connector types a consumer is exposed to (from dozens down to 3-5), it merely replaces old problems with new problems. -ethan

This is classicmp... Assuming by 'serial port' you mean somethign RS232-like (as opposed to USB, firewire (which are serial interfaces), HPIL, etc) then off th rtop of my head, Epson did (the RS232-interfacd drive units for the HX20/PX series of laptops). Resarch Machines did (disk unit for the 480Z, although that was a synchonous RS232 interface). Tandy/Radio Shcack (portable disk drive for the M100 etc). DEC did (TU58 tape cartridge drive). There must be many others. I am having problems following this. It appears you're saying that USB is superior to RS232 for all applications because USB is easier to use than, say, SCSI for linking up mass sotroage devices (but RS232 was not commonly used for this). -tony

I think most of those have been discussed on this list at some point :-) IIt's the only one I can think of off the top of my head, but I would hate to be dogmatic. However, at the time, hard disks were very expensive, and were thus only really used for staring large amounts of data which had to be accessible quickly (if the files were small enoguh to fit on floppies, just change floppies as needed :-)). And the throughput of an RS232 interface, even when set to 38400 baud (which was the fastest normally used for asynchonous transfers at the time) was not really high enough. So I would be slightly suprised if many RS232-interfaced hard disk units were made at that time. As I said earlier, I am exactly the opposite. And I think it comes from the fact that you use machines (in general) as the manufacturer intended, while I trie to do all sorts of odd things with mine. I don't doubt that if you plug a USB memory stick, or a USB-interfaced hard drive, into a modern machine where the OS has drivers for USB mass storage devies, then it works fine. But to be fair that's not unique to USB. I've never had any problems uysing HP's HPIN hard disk units with HP computers that support them. I've enver had any problems plgging DEC serial terminals into a PSP11. It's when you want to do things that hte manufacturer didn't intend, or had never thoguth od, that the 'fun' starts. And USB makes things harder than RE232 if you are doing things like that. A lot harder. -tony

On 7/20/10, Ray Arachelian <ray at arachelian.com> wrote: Yes. That's been a deficiency for some time. I've heard it works with ESX(i) 4.1. Unfortunately, we are using ESXi 3.5, ESXi 4.0, and ESX 4.0. :-( Probably, but in the enterprise world I work in, we moved away from the host-OS-based VM platforms (Server/Workstation) over a year ago, and except for losing our ability to plug in USB drives full of off-site backups, the change was a big win. We went from a Dell box that crashed and locked up every two weeks while running VMware Server 2 over Win2K8 to 10 months (and counting!) of continuous server uptime with ESXi 4.0 on the *same hardware*. That's another annoyance - plus if you plug in two USB serial dongles guess which device you were talking to the last time you ran. Considering there's a real-world device on the far end, consistency kinda matters. -ethan