Lost in Flash Drive PCI USB SCSI hotplug, etc.

Walt Pawley yellowdog-general@lists.terrasoftsolutions.com
Fri, 6 Aug 2004 22:19:05 -0700


On 8/6/04 9:32 PM -0300, Marcelo Giles wrote ...

>Congrats Walt! Good job.

Thank you but I don't feel good about it.

On 8/6/04 10:32 PM -0400, Andrew wrote ...

>> # mount -t msdos /dev/sda1 /mnt/flash
>
>You may put this in a bash script for quicker mount/umount:

True enough, but for now, I don't really need to be mucking with that.

Here's why, to both notions.

I started into this simply to test whether or not a combo FW/USB 2.0 card
would work with YDL. I'd previously mucked about with 3.0 so decided to try
my luck with 3.0.1 because the machine I was working with could handle it
... in theory. I wanted to do this because the card worked in Mac OS X
10.2.(3-6) with the vendor's software as a USB 2.0 speed device. Without
their driver, it only worked at 1.1 speeds. In 10.2.8, which is supposed to
handle such things without requiring an add-on driver, I could not get the
card to USB at all.

First, I don't feel good about how difficult it was to get a successful
install of YDL 3.0.1 even though the CDs fully check out as OK. Anaconda
kept coming up with unhandled exceptions until I chose to install a server
setup with add-ons. Apple's installer is far from perfect but things seem
to go smoother with installations of OS X. This is not all that surprising
as it's Unix masquerading as a nice, fuzzy, warm kitten of some species or
other so it doesn't need to satisfy the Unix maven with a plethora of
options beyond the ken of mere mortals.

>From my admittedly naive position, I can't help but believe that the Linux
installation process is just too fraught with difficulties to be handled by
almost anyone who sits a computer routinely. Getting a working foundation
installed should be a trivial pursuit, IMHO. Likewise, IMHO, the rest of
installation activities should work from that installed foundation. When
the process fails, it should not leave you starting from ground zero but
rather with the installed foundation and whatever got successfully
attached. I suppose this could be accomplished by a user by simply turning
off almost everything to be installed that they really want and then adding
them later.

Second, I've truly come to abhor computer documentation. I'm rather an old
fart who started working with computers when they were graduate academic
curiosities rather than default toys of the underprivileged. The people who
wrote documentation in those days were well lettered for the most part.
And, things were simpler. That has a huge affect on the nature of
documentation. People had time to create meaningful, complete, even
accurate descriptions of their work. Today, if you don't puzzle over the
uncommented C code ad nauseum, the English documentation often might as
well be Chinese for all the help it provides. Life's too short for that, so
software had better just plain work up front. Unfortunately, in the Unix
domain, it doesn't. I think it's part of the geek credo, or something.

Take, for example, the whole business of mapping Flash Drives to SCSI
logic. After stepping back a bit from intensely peering directly into
things, I found a bit of HTML in the docs that pointed this fact out ... in
passing. These same docs went on and on to tell about how I could do
specific things ... none of which I wanted to do. What I would have
preferred is some explanation of the logic of this arrangement. I suppose I
could blow a couple of grand and buy one of everything O'Reilly publishes -
but when would I have the time to read all that?

Third, I suppose I should be happy that "-t msdos" works. Despite my
elation at finally getting the Flash Drive mounted in some manner, I'm both
puzzled and disturbed by why the "-t hfs" nor "-t hfsplus" filesystem type
specifications don't work as at least one of them should. Or, at least,
that's what the docs seem to be saying.

I do believe the soap box I'm hopping about on is cracking, so I'll step
down now. Thanks for all the help!

-- 
Walter M. Pawley <walt@wump.org>
Wump Research & Company
676 River Bend Road, Roseburg, OR 97470
         541-672-8975