TL;DR - SCSI2SD isn't to blame for most of the problems people blame on it, which means a couple things:
- How do we better equip people who are trying to bootstrap a vintage mac "from nothing" using a scsi2sd or another tool. Some of this is a "what else can be built?" or "how can existing things be built to meet certain needs better" question and some of this is a "how can we better document what people can or should do to get started" question.
As a sub-point to this, a lot of problems people have with the SCSI2SD really do center around the fact that it was never possible to bootstrap and use a non-Apple drive from nothing when these machines were new. You always had to start from an existing system or from a disk utility or Mac OS install disc/kette equipped specifically to manage non-Apple drives. (Some, I'm think of APS specifically here, Mac drive vendors went as far as to include formatting software and preformat their drives with system 7 and an app or two pre-loaded, to get you started.)
- The SCSI2SD meets most needs, but if there's needs it doesn't meet, we should document what those are and present them to people who might be interested in building stuff. I understand that's what this thread is, but I also think that the proposal in this thread (an IDE adapter, specifically) is a bad idea for reasons I've outlined above.
For those of us who think a new SCSI to SATA adapter should exist, for example - why? What's the use case? who would use it? How much would it cost to build?
- Documentation for all of this stuff is spread to the four corners of the Internet and bringing some of it "home" (to the 68kMLA wiki) or at least aggregating links to what exists would be a good idea.
A lot of this comes down to each of us having figured things out a certain way and suggesting that way. I see this in most threads where people discuss the scsi2sd.
- A great example of this is that I see "lido" mentioned over and over but nobody has ever enumerated why lido in particular, among the dozen-or-so formatting utilities that exist. TBH I suspect it mostly has to do with whoever said it first and loudest.
We have this problem with a lot of things, but further discussion there is arguably for the wiki page/article ideas thread.
For the record, Maceffects is clearly a brilliant guy so if he’s having ongoing problems, it’s clearly possible to get really unlucky with these!
Unlucky and errors configuring are different things.
Based on the troubleshooting thread it seems like there are/were some configuration errors. As it stands,
we're still waiting on the latest in terms of what configurations are being applied to the SCSI2SD.
Getting unlucky is something that happens when you buy just one unit, or when one of several units is inexplicably nonfunctional or unstable, when others you have are, under the same configurations and conditions. When you're looking at three or more units that all behave similarly, I'm going to suspect that there's been a configuration error.
From my personal experience: using a third party formatting utility on an out-of-the-box SCSI2SD seems like the easiest way to avoid errors when copying configurations from screenshots or from tutorials, especially ones that, say, try to use the SCSI2SD to simulate a specific disk, such as the ST225 or whatever.
I'm
not that smart, all things considered, and I was able to dump a v6 into my 8600 and get it running mostly on my own, with a friend helping to do the firmware updates (which I recommend a lot) albeit with a little bit of troubleshooting.
- (I used a Zip disk to hold the utility I wanted to run, LaCie Silverlining, and a CD to install a fresh OS, so this is an option but I can see why people might want to just copy a system image to the card.)
(Incidentally it would make sense to think about building some OS install CD images that have these tools integrated, that's something for either someone else to do or me to do much later.)
(BTW SCSI2SD v6 performance on a fast PCI powermac, the second-fastest stock model pre-g3, is perfectly fine. I don't think it would do well at capturing video, but it'll do everything else I've ever wanted perfectly.)
I'm confident when it comes time to do it, configuring the v5 (which will probably start in my LC 520) will be most of the way as easy, give or take for the fact that the LC 520 doesn't have a built-in Zip drive, but there's ways around that.)
Point being - having more options would be great, but I find it suspect that one is
needed due to some issue with the SCSI2SD being fatally flawed somehow.
We should really be framing this as what we as a community or as vintage Mac users would get out of having another option available, in addition to the really good one we already have, rather than that the existing option isn't good, because I think it's fair to say we see on a regular basis, and I'll be unsurprised if maceffects' thread shakes down the same way, that most errors are because of mis-configuration, mis-mounting, or scsi bus and termination problems -- the last of which is a category of failures that impacts
anything that can be connected to a SCSI chain, not just the SCSI2SD.
To that end, I think that the best way forward would be to identify what the weak spots of the SCSI2SD v5 family and v6 are. It's well known that they don't meet every single need. What are those needs and what would help with them?
Most of the things the SCSI2SDs (or some variant of it) aren't already good at are, at least for the Mac, already being done by other options we have (PCI SATA/uscsi/IDE cards, floppyemu, a couple other things) but the problem in some of those cases is supply and cost. Not gonna lie here, a floppyemu
ready to go costs what three SCSI2SDs do (or two SCSI2SDs and two good SD cards and printed mounting/enclosure brackets.)
Personally, if I bought one, I'd prefer it to work as seamlessly as my FloppyEMU. I had no setup work at all. I just connect it, power on the SE/30, then make selections via the buttons and LCD. I would assume there's some formatting required via
LIDO for a SCSI2SD, but you've probably already done that and still have problems.
The reason there's so many valid setups is because the SCSI2SD is
extremely flexible. It was originally designed as a way to use flash memory for storage on a Mac LC, but it's evolved into a project that supports 68k Mac, PCs, PC98, MIDI synthesizers and samplers, UNIX workstations, VAX/Alpha,
probably old AS/400 machines, among other things.
It's also got functionality like presenting a big memory card as multiple devices, such as a CD/DVD-ROM drive, hard disks, and so on, again, for compatibility with different devices and to inexpensively add more than one type of function to a device. It's possible, for example, to image a system 7 install CD onto the scsi2sd and use that to install system 7 to a partition on the device,
for example.
The other thing this does is help with devices and platforms (Apple IIgs, very old Macs) where you don't really want to, say, show an entire 32-gig memory card to your old computer.
That and, of course, adding a screen makes things more costly and requires that configuration be limited to what three buttons and a 160x120 or whatever display can output. It works for the floppyemu because there's not a lot of configuration to be done, and because (unlike other Apple II floppy solutions like the CFFA) there isn't always a system control panel you can drop to to change media mid-task.
A straight SCSI-to-whatever converter is definitely possible, and the Artmix SCSI2SD clones (which violate GPL, apparently) have done away with a lot of the configurability in order to achieve simplicity, but you lose out on a lot of what makes the SCSI2SD flexible. What makes it so you can run a Mac Plus with it one day and a PowerMac G3 with it the next, an Amiga after that and a VAX after that.
I think there
is an argument to be had for direct SCSI-to-whatever converters, and the main argument that exists there, I'll argue, is for higher performance machines that can make full use of bigger and faster disks, but there's a big question of cost. The old Artmix CF adapters
ended their life a bit over what a SCSI2SD v6 now costs. Most of the SCSI2SD v6 variants cost around half that.
Is a hypothetical 8100/110 or 9150/120 owner who wants to get the fastest possible SCSI device going to be willing to pay potentially
a lot more (than a SCSI2SD v6, which costs $100) for a SCSI to SATA bridge, given that they have machines that can likely benefit from more performance but don't ahve PCI slots? (The same implicitly applies to anything else mentioned or where the SCSI2SD
can be used but where there's not a viable way to add SATA.)
It's tough because you either need to get something that lets you modernize/replace/augment other functionality (SATA DVD or blu-ray burner for DVDRAM functionality and added CD-ROM reliability?) or has a meaningful performance benefit while making some other part of the system better available for other things (such as if the stock "build a bad powermac g4" build-out for six-slot powermacs didn't already have a free slot
after a SATA/IDE/SCSI add-in card, or if you wanted to build an actual Avid 9600 or video capture/DV GV using a SCSI add-in card instead of a SATA one.)
The problem is none of those things are really beneficial to anyone with anything less than an 040. A SATA adapter would honestly be bad in that scenario because pre-040 your ability to run anything over than 2-4GB volumes is very, very compromised in a lot of ways. Though, it's a little entertaining to imagine someone slamming an 80-gig laptop hard disk into an SE/30 only to run like three 2GB partitions on it. (I can't talk that much: my IIgs has a 32-gig CF card with 8*32-meg partitions, or 256 entire megs being provisioned.)
--------------------------
Back to "more generally" a problem I see with the SCSI2SD is that it often takes the blame when, just, someone doesn't have a lot of infrastructure.
I talk about "the advantage" of being able to use my 8600's Zip drive to help run partitioning software, but tools like that have been common for the entire time the Mac as a platform has existed. You needed a diskette with Lido or Silverlining on it if you had
any third party hard disks on your macs literally up until USB was a thing
in 1998. (And: LaCie built a USB hard disk and bundled Silverlining with it
anyway even after that.)
So, like, in that sense a lot of our problem is that people are picking up bare machines from garage sales and they want to get started with
literally nothing and that has never been practical or easy. So, again, instead of vilifying the SCSI2SD for not being a one-stop solution for that, we really need to ask ourselves what we can do to make that kind of set-up easier.
It's not entirely out of the question that getting that bootstrap infrastructure isn't something the SCSI2SD could do, the issue of course would be communicating the need to aperezbios/inertialcomputing, the design, the build, and then telling people what one would meet their needs, and documenting it.
Of course, other tools that meet this need like floppyemu
also exist, and there's techniques such as the one Crutch mentioned where you can use a modern computer to write a pre-built disk image onto the scsi2sd and then add more from there. (There's also the idea of selling/distributing pre-configured SD cards for the scsi2sd v5 family, something maybe we could do but maybe inertialcomputing couldn't or wouldn't want to do, for liability reasons.)