Date: Mon, 12 May 1997 14:18:52 +1200 From: physmsa AT cantua DOT canterbury DOT ac DOT nz (Mr M S Aitchison) Subject: Re: Back on track... Opendos's Not Unix! To: goehrigd AT gort DOT canisius DOT edu, opendos AT delorie DOT com Message-id: <199705120218.OAA03271@cantua.canterbury.ac.nz> Precedence: bulk > From goehrigd AT gort DOT canisius DOT edu Fri May 9 23:26:46 1997 > > Maybe the solution is to just create & bundle in: > LinuxEMU, a pseudo-UNIX envirionment that allows you to... > ... Why bother running OD at all.. Answer? Opendos is Smaller... Well, I think we both don't want DOS to become Linux, and want to retain DOS advantages in OpenDOS. But if it isn't Linux, what is it? Somebody (perhaps all of us, perhaps only Caldera) have to settle on at least a broad idea of what OpenDOS is all about, what it is trying to be. The trouble, which I tried to convey in earlier messages, is that OpenDOS is not organised as well as (say) Linux or FreeDOS - it isn't even clear what is something "we" can decide and what is entirely up to Caldera. I'd like a set of design guidelines drawn up, like the following (but the actual details are up for debate)... OBJECTIVE 1: Whatever special versions of OpenDOS might come along, there will always be a plain "vanilla" version that runs on low-end computers such as an original IBM 256Kb RAM PC. OBJECTIVE 2: OpenDOS will continue to run old DOS applications; it will be about as compatible with PC-DOS/MS-DOS as one version of MSDOS is with another. OBJECTIVE 3: The setup and on-going administration will be as easy as possible, and the user interface both ergonomic and compatibile with traditional COMMAND.COM (if there is any conflict between convenience and compatibility, the user should get the option). OBJECTIVE 4: Security (against viruses, access to private data, and ability to restore a working system) should be as good as reasonably possible. This means, at the least, restoring access rights available under the old Multiuser DRDOS, but it could go a lot further (especialy virus resistance, not just detection). OBJECTIVE 5: Provision of modern conveniences, such as long filenames, object-orientation, web browsers. It should be easier to port sources from other systems to OpenDOS than plain DOS, perhaps in the way that EMX makes it easy to port Unix sources to OS/2. OpenDOS should not only "keep up with the Jones", it should be able to get features before most commercial counterparts. OBJECTIVE 6: OpenDOS should be the most inter-operable DOS; it should be happy with Mac/VMS or Unix text files (i.e. not CR-LF), be able to use industry-standard printing and file sharing systems without headaches (e.g. should work with Unix-style permissions on mounted file systems; use distributed configuration systems like NIS, DNS, NDS). It should be able to use X11, and even if a low-RAM system is using a non-X11 standard GUI, the system should be similar enough to administer. OBJECTIVE 7: When it comes to high-performance 32-bit facilities and SMP, it might be best to not try to extend OpenDOS to do do things Linux has solved already, but to run OpenDOS within Linux "seemlessly". To achieve most objectives, it is important to have good fundamental design decisions early on, like a flexible efficient IFS, and a good security model (anybody aware of how bad MS's is?). The last two objectives are probably the most controversial. They might sound like I'm trying to make OpenDOS into Linux. Notice that I'm not saying OpenDOS should look and feel like Linux... traditional DOS users should find it as easy to use as ever. But developers can get a lot of benefit by not trying to re-invent the wheel - if possible, the good work that has already been done in Linux should be usable if we are careful with compatiblity. Ultimately a 32-bit version of OpenDOS will be great, but I think we need to get the present version running really well first, and let it work on top of Linux for the benefit of both those running Linux who want DOS as well as those wanting powerful features like SMP support and special applications such as firewalls for DOS. Also, people running large numbers of computers will appreciate support for industry standards such as DNS and SNMP; again this need not impact on the way OpenDOS looks and feels to the end-user. Even those who aren't interested in Unix should be able to get considerable advantages out of the *good* parts of Unix system administration ideas. ------------------------------------------------------------------------------- Mark Aitchison, Physics & Astronomy \_ Phone : +64 3 3642-947 a.h. 3371-225 University of Canterbury, (/' -------------------------------------------------------------------------------