Date: Wed, 28 Nov 2001 23:53:22 -0600 From: Mike Meyer <mwm@mired.org> To: "Anthony Atkielski" <anthony@freebie.atkielski.com> Cc: "Andrew C. Hornback" <achornback@worldnet.att.net>, <chat@freebsd.org> Subject: Re: Feeding the Troll (Was: freebsd as a desktop ?) Message-ID: <15365.52562.394957.602907@guru.mired.org> In-Reply-To: <017f01c1788c$8cb71d90$0a00000a@atkielski.com> References: <004801c17872$98e47b40$6600000a@ach.domain> <017f01c1788c$8cb71d90$0a00000a@atkielski.com>
next in thread | previous in thread | raw e-mail | index | archive | help
[This thread is being globally redirected to -chat, along with the prior one. Others please do the same.] Anthony Atkielski <anthony@freebie.atkielski.com> types: > Andrew writes: > > So, both Sun Microsystems and Silicon Graphics > > (among others) were wrong to use a Unix-type OS > > for their high end GRAPHICAL workstations? > In what sense? It was certainly a poor technical decision. However, writing a > new operating system costs money, and so does buying a new operating system from > someone else. UNIX was at hand and about as close to open source and free as > one could get. > > That is also why Apple chose UNIX as a basis for Mac OS X. They couldn't afford > to write something new, so they reused as much as they could, even though this > is not necessarily a good idea from a technical standpoint. Note that Apple examined an OS that didn't have all the multi-user complexity that Unix has, ran - like a bat out of hell - on Apple hardware, and on MP boxes with those processors. They decided against adopting it. > Microsoft did the same with Windows 3.x and (to a limited extent) Windows 95. > Windows NT had elements of OS/2 architecture in it as well. Not to mention VMS. > Writing a new operating system costs billions of dollars; even Microsoft cannot > afford to do that. No, it doesn't cost billions of dollars. Sun, Apple and DEC have all done OS development projects that they abandoned. I'm pretty sure Be didn't have billions of dollars to spend, and they put a very nice little OS on the market, though it wasn't suitable for much off the desktop. > > I'm sure Scott (Mr. McNealy) is crying over that > > comment. > It's possible to make lots of money with technically less-than-ideal products. Gee, I'd say that MicroSoft's products are the best example of that. In fact, I've even written a paper on the topic inspired by a chapter in Richard Gabriel's "Good News/Bad News/How to Win Big". That chapter discussed why Unix was beating *his* favorite working environment out in the market, in spite of it being a clearly superior environment - a sentiment which I agree with, BTW. His chapter was titled "Worse is better". My paper is titled "Good enough is best", and can be found at <URL: http://www.mired.org/home/mwm/good-enough.html >. > Sun and SGI weren't selling graphics workstations on the basis of the underlying > OS. Of course not - the underlying OS is largely irrelevant if it provides all the needed functions. Any functionality beyond that is also irrelevant, except that it might chew up resources that could be better used elsewhere. > > Multiple independent processes... the sort of > > thing that you'd want to be doing if you had > > multiple instances of a data analysis program > > running. > No. For data analysis, you need programs that communicate extensively, not > completely independent processes. Horse puckey. If you've used batch environments, you might be familiar with SAS, which communicated large volumes of data, but did it by passing a data set (or a small number of them) from one command to the next. While it wasn't as popular as SPSS in those environments, it was much easier to use. I even did some of my data analsys coursework using it. This model maps very well onto Unix, and can be found implemented in the math/unixstat port. Some environments *do* require lots of communications. Unix works fine for that. Some of the best tools for that I've run into are programs that bundle the data up with C code to manipulate it, then make the whole available as objects in Python. The GUI is done in TkInter, meaning the code can be run on Unix, Windows, or the Mac with little or no modification. > > Which makes it perfect for Unix. > See above. UNIX (and most other operating systems) lack the close communication > between processes that would be optimal for this type of application. Most - if not all - modern Unices allow multiple independent threads of control in a single address space. If there's a closer form of communication in a computing environment, I'm not familiar with it. > > ... I understand that you're coming from a closed= > > minded Microsoft point of view... > No. I come from a non-religious point of view, and when one is not encumbered > by religious faith in a particular platform or operating system, one tends to > see advantages and disadvantages of each system more clearly. I do not worship > a FreeBSD god (nor a Windows god), so I do not feel frightened by the idea of > running two completely different operating systems for two entirely different > purposes. I choose the tool that fits the job. I'm not blinded by a religious faith in any particular platform either. For years, I ran *three* different computing environments - one on my desktop, one on my clients servers, and one in my pocket, and that's counting the many versions of Unix they ran as one platform, which isn't quite true. I think you're blinded by yourself, and are assuming that because some tool fits you best for a particular job, it must do the same for everyone. That's not true. I recommend that most of my relatives run Windows, not FreeBSD. Before Windows was popular, I recommended that they run DOS. With those who grouse about Windows being inferior to the desktop environment they had before they drank the koolaide, I'll discuss what they want and possibly recommend FreeBSD if I think it will fit the bill. I might recommend a Linux distro instead, if I think that that would fit them better than FreeBSD. > > ... but, until you have used and properly evaluated > > FreeBSD (or any other Unix, for that matter) in > > a desktop environment, I don't believe you have > > the right to say that it makes a poor desktop environment. > I knew within a few days of installing FreeBSD that it was suboptimal as a > desktop (at best). But you still haven't given us a description of *what* makes it suboptimal. You say that it's architecture isn't adequate, but that's a "why", not a "what". Come on - what do you want to do on your desktop that you can't do on FreeBSD? I'll grant that Windows is much better at working with files in proprietary MicroSoft formats, but that doesn't make it a better desktop environment, that makes it a better environment for exchanging files in those formats with other people. Those are different things, and not everyone needs that particular ability in their desktop. > > Making assertions without examples, proof or fact > > to cite to back them up is not good. > I've explained my reasoning in considerable detail. To those for whom IT is not > a religion, my explanations are cogent. For the true believers, no "proof" is > ever adequate to sway them from the Path. I've also explained why Unix makes a perfectly adequate desktop for the heavy user in considerable detail. To those for whom the desktop is not a religion, my explanations are cogent and sufficient. For true believers, no "proof" is ever adequate to sway them from The One True Path. I have as yet to see you list a single thing you wanted FreeBSD to do as a desktop that it couldn't. > > At one time, you could run Windows on everything > > from a Macintosh to the latest and greatest > > from Intel to the wicked fast Alpha from DEC. For a real giggle, consider that I've seen Windows running on the MIPS processor running with the opposite endianness of the MIPS boxes. DEC killed that product when the dropped the hardware line in favor of the Alpha. > > So, from what you're saying here is that you > > advocate one operating system for each type of > > application? That would be absurd. > What's absurd about it? In fact, that's exactly how real-world implementations > have typically done it, in the absence of political reasons for forcing one > "solution" to fit all environments. Many organizations use Windows on desktops > but UNIX on servers. Yup. Many also use Unix on the desktop along with Windows. And many organizations use Windows as servers, and are quite happy doing so with no political persuasion at all. So what? If something works for someone, I'm not going to tell them it's not adequate to do the job they've got it doing. Only an idiot would do something like that. > With stable applications, yes. With poorly-written applications (the norm in > PC-land, alas!), they do not stand up very well. Poorly-written applications > are best run on NT, if possible, because NT will not crash when they do. The same was true of OS/2. I know a lot of people who developed for Windows on OS/2 for just that reason. Look at all the good that did IBM. <mike -- Mike Meyer <mwm@mired.org> http://www.mired.org/home/mwm/ Q: How do you make the gods laugh? A: Tell them your plans. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-chat" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?15365.52562.394957.602907>