OSnux or Li/2?
Recently in an IRC group: want to write an editorial for OS2.org? Well, dunno, and if so, I don't have the nerve to write the N+1st successor of the "OS/2 isn't dead" story. Maybe "OS/2 goes Unix?" Now, there has been recently a mail conversation with me leading to an article by someone who BTW didn't mention my contribution to it too excessively, I do consider this still a topic. On the other hand, is it relevant? Let's see.
So: quo vadis, OS/2?
First of all, fact is the IBM is going to position OS/2 as "Aurora" in the server domain. The desktop, which "us" rank and file is interested in and which we cannot just lay a four-place amount on the counter for, is ruled by this weird, not even Y2K compliant what's-its-name (which admittedly I also have on my disk, for the situation I feel I have to run after Lara Croft, in other words, a toy system). At least, IBM is still eager to deliver fixpacks for Warp 4 (that FP31 is now end-of-life for Warp 3 is in my opinion no reason for mourning - to whom OS/2 is worth something *has* Warp 4 - at some time one should finish the attitude that even an OS/2 flogged for 5USD is by far too expensive - even for a low student's budget; good tools are expensive and worth its money - garbage tools are bundled as shovelware with a newly bought PC).
Certainly, we'd like to have WSeB, not for the server or for e-business, but because of the much-praised features like JFS (actually rather buggy, with fix #1 recently arrived), the SMP support (but who of us really has a dual board, and even worse: *who* actually has the applications that really take advantage of it?), and some more hidden internals like the KEE (32bit device helper API) and the 32 bit IFS API (that's rumour, there is no such beast anywhere in the 14.039SMP kernel). Anyway, in my opinion these are details attacting at most the freaks, but nothing to really help the OS/2 community to get on?
Leading to the question: what might be helpful? OS/2 is and has always been the crossbar switch for compatibility. Until win95 arrived with its apparently intentionally incompatible API (remember the Win32s > 1.25 incident?), DOS and Win3.x were known to run better in OS/2's DOS box than in native DOS (at least, it didn't crash the whole PC when it was periodicall time for an UAE). XFree86/OS2 which I confess to be guilty for, adds connectivity to Unix hosts in the network. Activities like Everblue and Odin are well alive and will some day in the same style (namely compatibility) support OS/2. Lovely Lara under OS/2? Well possible...
But I am not very happy about this. Why? The activities of the freeware and Open Source community are touching, but we can't neglect the problem with hardware support in the foreseeable future - some claim that\'s already the case now; I don't: there *is* support now, but again, admittedly you can't take your hardware, like with the what's-its-name, from the retailer's 5$ garbage cans.
The current activities, like Odin, supposedly to give us Win32 compatibility, are aiming *towards* OS/2. My idea goes into the other direction: if users won't come to OS/2, then give them OS/2 through their backdoor. IBM has obviously abandoned us, so why should we improve their system further? What really matter is us, and the OS/2 applications we all like so much. Imagine there were a WPS for Linux: how many people would then change immediately? Supposedly all those who have understood and like object centered usage of OS/2 and its general stability. Imagine further Linux or also FreeBSD had a loader that would load and execute native OS/2 applications? It is already reality nowadays that the x86 Unices have an almost any-to-any binary compatibility: provided you have the right runtime libraries, you can run BSD binaries on Linux, Linux binaries on Solaris, SCO code under BSD etc. Now we'd have an operating system that allows us to work with the same software we're used to for long, with the same look&feel as before, but below the surface, there is a Unix kernel. Additionally, since there is ongoing work with this system, we are not in danger to lose hardware support. Linux runs on Merced and Alpha, and really supports most of the recent hardware you can get access to.
Admittedly, some anachronisms will be different: there isn't very likely going to be a CONFIG.SYS, nor will there be drive letters (but then, who of us is *not* using TVFS, anyway?), and instead of MPTS, we'll start YAST or RPM, if we feel the necessity to tinker with the network configuration, but then, even IBM has surprised us with new tools for administration in Aurora (like FDISK replaced with LVM). We're going to say good-bye to Watchcat and Theseus as well, as these tools are so deeply interwoven with the kernel that an emulation of them might result in a rewrite of the OS2KRNL. I have no problems to forget about those special purpose software.
However, the idea of a Unix/2 or a libdoscalls.so.4.0 will remain a dream if noone is going to make it become reality. The knowhow is available worldwide, but it is no job for a split group of three or four engaged people.
On the other hand, the job is not impossible. I always hear this excuse from people when asked to fiddle a little bit with XFree86/OS2 code: "Nah, you have to know so much (about X11 or PM or OS/2), and I only have some experience in C, and I have never written such an OS/2 device driver like xf86sup.sys." Let me tell you a secret: before XF86SUP.SYS, nor did I. You are growing with your tasks. There is much to learn. But it works.
Regards/2
Holger Veit, 1999-09-02
Comments: 4
|
![[Forum]](../../../images.os2.org/en/forum.gif){kind=small}
![[Forum]](../../../images.os2.org/en/forum2.gif){kind=small}
