Guess who I've had on the phone (Friday night, this is)...? Bob Brenchley!
We talked for about an hour, and it seems that I've done him some injustices (well, he did them to himself, and is now contradicting himself, but never mind)... It appears that the plan is for any IDE drive to be used, but for <60Mb ones to be released with the interface at first to get them out for less than L150. Anyway, he was phoning me up about the HD interface and DOS system... Apparently the partition table idea is a no-no... this is what him and Nev Young have come up with... HDOS is their system, which *everything* has to go through to get access to the HD (it's "naughty" to access the IDE drive direct). In other words, it's a memory resident level between hardware and software (including CP/M, Unix, et al), which accepts logical sector numbers and spits out read/writes to the appropriate sector on the drive. To create a "partition", you allocate a file using HDOS (say 20Mb long), which you then access using logical sectors which HDOS references for you... the advantage of this being totally flexible partition size (not quite... I can see problems with fixed-at-setup data structures etc which rely on the partition being a given size forever) which doesn't have to be contiguous. As far as the DOS is concerned, some routines will be kept on the HD itself (COPY, FORMAT are both big contenders for this), and there will be a number of overlay utilities which can be loaded into the 16/32k DOS page for this purpose... There will be something "like the Windows Swap file system" which will allow HDOS to store any tables etc it generates on the HD for later referencing (I think we're talking about Temp files here folks). When I suggested making sure that if the DOS has to be bigger than 16k (very likely) that the core routines which interface with BASIC etc are written to run from section B of the memory map (16384-32767), and others are written to run from 32768-49151 (so that they could, if the user has the memory) be moved into external memory, Bob said that if people could buy a 60Mb hard drive for L150, they won't want to buy 1Mb of memory for L90. (Which is why Martin's been working along the lines of a SIMMS memory interface for the SAM ! ) SD software are likely contenders to release the HD first, then West Coast will release a SAM-with-hd later... Erm... let's see... The interface will not be sold separately for at least 6months -- IDE isn't kept "standard" enough apparently, so to avoid problems with slightly dodgy drives a "send us your drive, we'll make sure it works and then send it back to you with interface" policy will be used. Cost cutting will _possibly_ be done by providing the +5V side of the drive from the SAM and the IDE interface, and the 12V supply from an off-the-shelf 12Vtransformer + regulator at 700mA in a 3-pin plug... ((Martin and I have serious doubts about this -- we need 12V at 1A to power the drive we've got upstairs (a western digital caviar 280), so I think there will be probs...same goes for getting the +5V from the interface/SAM)) Bob's still insistent on this +D syntax crap too... but the ROM which handles it (and some of the HDOS stuff) will be released as cheap as possible -- L400 inc p&p if poss (we doubt it!!!!!!!!) Anyway... back to the land of the living... what do you think guys? Bob doesn't want there to be a split in the SAM community, so this is all being discussed at length, and he's trying to get everyone to agree on one standard.. currently, I can see appealing aspects of his system, but I prefer my partition table idea, as it's more program-specific if it needs to be... Any takers? Si
