On <03 Nov 94 10:38> [EMAIL PROTECTED] wrote: Cs> Yeah - I hate the way MS-DOS does Cs> that. It's not hard to look at the Cs> contents of the FAT to work out the Cs> number of free clusters, and Cs> *then* start to write the file, not Cs> the other way around. Some Cs> people, eh?
That's right! Or in the situation when an app Knows the final size of its output file, WHY don't people do a random-write with zero-fill of previously unallocated sectors the entire output file... If an error ocours then just prompt for a new drive spec etc before true output starts... plus by creating the file as a blank sheet it'll be properly allocated and can be read whilst still being written to later on (For ZIP204 and RAR decompression without a massive dictionary buffer) Cs> It'll be more virtualised -- if you Cs> link paritions (in the final Cs> DOS), then they will appear to be the Cs> same -- you will be able to Cs> allocate partitions as subdirectories Cs> and vice versa. I'm thinking of Cs> implementing disk images -- you copy a Cs> SAM disk in, and tell the DOS Cs> to map it to drive 1 or drive 2 -- Cs> then you can do all the sector Cs> access as normal :) Is there a team of coders laying down the code or are you to be churning this all out on your own? If it works I'll be very impressed! Be prepared though to compromise, it's better than being forced to give-up due to overwhelming complexity part way through just to avoid admitting that some luxuries arn't worth the coding involved, remember that most modern OS's that do the things you envisage are either written in a HLL compiled language (NT,unix) or have teams of specialist well paid coders working exclusivly on them or both! Cs> It's the former -- LBA mode, set as Cs> part of the features register, Cs> allows you to use 24bit (I think -- Cs> not got the specs in front of me) Cs> logical sector numbers to address the Cs> hard disk. Will make it a *LOT* Cs> easier, and also will make it a piece Cs> of cake to convert to SCSI or Cs> even to an IDE CD Rom drive using its Cs> seconds based address scheme. I'll have get that ATA doc over on this machine and give it bloody good read.. Plus there's a posibility that some early PeeCee oriented drives might not implement that feature! Also it is reported that not all IDE drives support free format tanslated translates, some only do the ones required to get a peecee-clone talking to it and ignore all other translate requests:-( Partition table will be in cylinders though, won't it? I hope so anyway:-) Cheers. Johnathan. ___ Olms 1.60 [Evaluation] +-------------------------------------------------------------------+ | Standard disclaimer: The views of this user are strictly his own | | ===> Gated @ Centronics BBS [centron.com] +44-1473-273246] <=== | +-------------------------------------------------------------------+
