| boltar@nowhere.co.uk: Dec 11 09:49AM On Tue, 10 Dec 2019 21:09:24 +0000 >After using Windows 3.1 (the 16 bit one) Windows NT 3.1 was a breath of >fresh air. It just kept running. It really didn't crash (except when our >HW was broken, and since we were developing it that did happen). Perhaps I was spoilt - I went straight from 8 bit at home to unix at university so I was used to using HP-UX and Solaris at 21. Then I was introduced to Windows. It was like going from a Ferrari to a Lada and NT didn't nothing to dispel that imagine. It crashed, it had to rebooted once a week as it had a serious memory leak issues and it was a slow as a dog. If that was a breath of fresh air then god knows what you were breathing beforehand. |
| James Kuyper <jameskuyper@alumni.caltech.edu>: Dec 11 03:13AM -0800 On Wednesday, December 11, 2019 at 4:50:06 AM UTC-5, bol...@nowhere.co.uk wrote: ... > dispel that imagine. It crashed, it had to rebooted once a week as it had > a serious memory leak issues and it was a slow as a dog. If that was a breath > of fresh air then god knows what you were breathing beforehand. As I remember it, NT almost never crashed, but I'm not sure what I meant by "almost never" when I formed that memory. Windows 3.1 typically crashed several times per day for me, so a weekly crash might have felt like "almost never" by comparison. Like you, my experience with Unix pre-dated my experience with either DOS or Windows, and those Unix systems were very reliable. However, those systems were mainframes being maintained by professional sysadmins; I would not have dreamed of expecting the same reliability from a desktop machine that I maintained myself. |
| boltar@nowhere.co.uk: Dec 11 11:20AM On Wed, 11 Dec 2019 03:13:44 -0800 (PST) >those systems were mainframes being maintained by professional >sysadmins; I would not have dreamed of expecting the same reliability >from a desktop machine that I maintained myself. The basic reliability of an OS should be independant of how well or otherwise its administered so longs as patches are installed. |
| "Fred. Zwarts" <F.Zwarts@KVI.nl>: Dec 11 12:57PM +0100 >>from a desktop machine that I maintained myself. > The basic reliability of an OS should be independant of how well or otherwise > its administered so longs as patches are installed. The reliability also depends on the amount of (third-party, cheap) hardware used in those systems. Those early Unix systems supported far less hardware. There is a reason why Microsoft later introduced Microsoft certified drivers. I was used to VMS systems. I remembered how in that time the same argument was used against Unix and pro VMS. When later-on Linux started to supported the same hardware as Windows, I also saw "kernel panic" very often. (But it was not blue. :-) |
| Bart <bc@freeuk.com>: Dec 11 12:11PM > Perhaps I was spoilt - I went straight from 8 bit at home to unix at university > so I was used to using HP-UX and Solaris at 21. Then I was introduced to > Windows. It was like going from a Ferrari to a Lada My experience was the opposite. I didn't think a great deal of Windows, which to my mind simply got in the way (compared to my previously using MS/PC-DOS for ten years where you had direct control of a lot more of the hardware). But trying out Linux every few years, Windows always looked and felt solidly professional by comparison. On Windows, things just worked (mouse, USB, printers, internet, display), whereas on Linux during the 90s, these rarely worked properly, or worked badly, or had to be coaxed into working via configuration files. And even when the display worked, it looked dreadful, like a cheap 80s home computer. That's probably not surprising considering Windows was backed by a multi-billion dollar company, and supported by multitudinous hardware vendors, while Linux was hacked together by amateurs in their spare time, it seemed. (And with dozens of different distributions.) > so I was used to using HP-UX and Solaris at 21. If you were writing software for use by the general public, then Windows machines were obtainable anywhere at any PC store, all ready to go. That was a huge advantage compared to having software relying on obscure workstations (where did you even buy such a thing). I suspect such machines were more expensive too. |
| boltar@nowhere.co.uk: Dec 11 12:42PM On Wed, 11 Dec 2019 12:11:33 +0000 >the hardware). >But trying out Linux every few years, Windows always looked and felt >solidly professional by comparison. On Windows, things just worked Well everyones experience is different I suppose. I've run Linux since kernel version 1.1 and most things worked fine apart from printers which have always been its weak point. The Windows PCs I was using at work however couldn't last more than 2 or 3 days without a reboot if they didn't first crash anyway. >And even when the display worked, it looked dreadful, like a cheap 80s >home computer. That's probably not surprising considering Windows was Not in my experience. Yes, X could be a pig to get working but once it was it was fine. In fact I had full motion video running in linux on a 486 back in the 90s. Something Windows wadsn't even close to achieving except with special 3rd party hardware. |
| David Brown <david.brown@hesbynett.no>: Dec 11 04:01PM +0100 On 11/12/2019 13:11, Bart wrote: > (mouse, USB, printers, internet, display), whereas on Linux during the > 90s, these rarely worked properly, or worked badly, or had to be coaxed > into working via configuration files. There is a /huge/ difference between HP-UX, Solaris, Sun-OS, AIX, etc., and Linux - especially at the time in question. Linux these days is strongly supported by big companies in all senses (big companies use it, big companies supply it, big companies provide support, big companies make sure their hardware and software works with Linux, etc.). But in the days of Win 3.1 and NT 3.x, Linux was very much a "do it yourself" system for people who were willing to get their hands dirty. The big unixes were a different world, however. You did not buy a random "PC compatible" and try to run Solaris on it - you got a Solaris workstation, such as SPARCStation, and used that. Often, you used it on a network - a serious Ethernet network with a proper server, not a little "LANtastic" setup. The hardware was a world ahead of the PC world in terms of quality, speed, and features - and a different world in terms of price. I went from mainly 8-bit systems (Spectrum and BBC micro), to SunOS then Solaris at university, and then to Windows 3.1. I have to agree with Boltar - going to Windows after Solaris was huge leap backwards in both hardware and software. The SPARCStations had vastly better screens and graphics than anything you got in the PC world - and if you needed more screen space, you logged into the computer next to you and treated its screen as an extension to your own. (The network transparency of X might make it poor for 3D graphics and videos, but was excellent for that kind of thing.) Virtual desktops were standard in the *nix world over 30 years ago, and only with Windows 10 is there a limited version in the Windows world. The Windows world has never been close to the Unix world in terms of quality, solidity, or professionalism. > That was a huge advantage compared to having software relying on obscure > workstations (where did you even buy such a thing). > I suspect such machines were more expensive too. Of course they were more expensive. They were worth more, and did more. But they were used by companies (and universities) willing to pay for their tools, rather than those that wanted the cheapest machines they could get and didn't care how much time and effort was wasted with reboots, reinstalls, crashes, crappy hardware, crappy software, etc., - or by companies that knew workstations were better but couldn't afford them, or when they needed hardware or software that only existed in the Windows world. Anyone with the choice would have gone for a Unix workstation over Windows. |
| Vir Campestris <vir.campestris@invalid.invalid>: Dec 10 09:09PM > Are you talking about NT here or some reliable and fast OS that MS never > released? Win NT was a blue screening, slow joke of an OS and frankly I'd have > expected better from people who worked on VMS. After using Windows 3.1 (the 16 bit one) Windows NT 3.1 was a breath of fresh air. It just kept running. It really didn't crash (except when our HW was broken, and since we were developing it that did happen). 3.51 was if anything better. AIUI NT4 picked up the 9x GUI. Certainly it was not as stable as the earlier ones. Andy |
| boltar@nowhere.co.uk: Dec 11 04:15PM On Wed, 11 Dec 2019 16:01:32 +0100 >that kind of thing.) Virtual desktops were standard in the *nix world >over 30 years ago, and only with Windows 10 is there a limited version >in the Windows world. Indeed. Such a shame though that Sun management had such myopia and didn't see both the PC and Apple creaping up on them as they treaded water with their hardware in the 00s, but instead kept charging silly money for their workstations instead of pitching the price at Apple levels where they could have shared the professional graphics market. Instead Apple (and to a lesser extent Windows) ate their lunch, ironically with their own version of Unix in the end. Plus they didn't really see Linux coming in the server space until it was too late. |
| "Öö Tiib" <ootiib@hot.ee>: Dec 11 12:48PM -0800 On Tuesday, 10 December 2019 23:56:09 UTC+2, Daniel wrote: > > also wants to have some garbage-free programming language. > As opposed to a garbage-full programming language like C++? :-) Or did you > mean as opposed to a language with GC? But Go has GC. Yes I meant as opposed to historically garbage-filled C++ and failed attempts of reducing that garbage like Java and C#. C++ is useful only by having "The Bible Of Its Usage In This Project". A sketch of it: <https://isocpp.github.io/CppCoreGuidelines/CppCoreGuidelines>. Basically one without tens of millions dollars to throw into project has to limit allowable C++ constructs in it to bare minimum right from the very start or to do it in next iteration after failing the first attempt. > > The syntax of Rust is not that different from Swift or Go. > The memory ownership models are different. Go has GC. Swift has > reference counting. Rust has affine types. The GC does not worry me, "garbage collection" is just unfortunate name to great idea of automated memory recycling. Its only issue is that it makes programmers to feel safe while it totally failed to automate recycling of all other, far more valuable resources than memory. Plus it does not favor data layouts that are naturally suitable for parallel processing. Other nice GC languages besides Go are JavaSript (non-tameable wild) and Kotlin (sponsored by Jetbrains and Google). > > are no such "gotcha!" bear traps in Rust like C++ is full of > > and so it perhaps suits novices better than C++. > I'd be interested to know what you think about Rust's affine types. I'm currently on chapter 4 of The Rust Programming Language, The Stack and the Heap, and have some mental resistance to "_don't_ do as the ints do". I think of those as of policy that only owner has right to use and ownership is move-only. I basically follow same policy in all other languages (but it is easiest to automate in C++). I basically allow shared ownership of immutable objects as the only exception. |
| "Chris M. Thomasson" <chris.m.thomasson.1@gmail.com>: Dec 11 02:17PM -0800 On 12/11/2019 3:13 AM, James Kuyper wrote: >> a serious memory leak issues and it was a slow as a dog. If that was a breath >> of fresh air then god knows what you were breathing beforehand. > As I remember it, NT almost never crashed, One time I created a little program that would try to crash NT using a build up of non-pageable memory. It took mem from the non-paged pool, iirc. This can be done using IOCP. Each pending request is in non-pageable memory. The result was a crashed system. This can happen in a server using IOCP, its a real problem indeed. So, during server setup, my test program tried to stress the system to certain breaking point(s), then setup a configuration file that had the limits hardcoded. |
| You received this digest because you're subscribed to updates for this group. You can change your settings on the group membership page. To unsubscribe from this group and stop receiving emails from it send an email to comp.lang.c+++unsubscribe@googlegroups.com. |
No comments:
Post a Comment