which was the best X11(R4)-terminal?

Additional operating system/hardware discussion (Windows, Linux, *BSD and others)
Forum rules
Any posts concerning pirated software or offering to buy/sell/trade commercial software are subject to removal.
User avatar
miod
Posts: 477
Joined: Fri Oct 09, 2009 2:44 am
Location: Clermont-Ferrand (France)
Contact:

Re: which was the best X11(R4)-terminal?

Unread postby miod » Wed Apr 19, 2017 1:30 pm

Y888099 wrote:Never seen one. According to its manuals it's a superscalar RISC. I assume it's faster than 68k. But I wonder how comfortable was the software if there wasn't WxWorks (not existing for 88k, right?) under the hood.

I'd rather consider the lack of this piece of crap being ported, a good sign, myself :mrgreen:
But I don't have an objective point of view when it comes to m88k (I do when it comes to VxShitWorks).
:Indigo:R4000 :Indigo:R4000 :Indigo:R4000 :Indigo2:R4400 :Indigo2IMP:R4400 :Indigo2:R8000 :Indigo2IMP:R10000 :Indy:R4000PC :Indy:R4000SC :Indy:R4600 :Indy:R5000SC :O2:R5000 :O2:RM7000 :Octane:2xR10000 :Octane:R12000 :O200:2xR12000 :O200: - :O200:2x2xR10000 :Fuel:R16000 :O3x0:4xR16000 :A350:
among more than 150 machines : Apollo, Data General, Digital, HP, IBM, MIPS before SGI, Motorola, NeXT, SGI, Solbourne, Sun...

Y888099
Posts: 393
Joined: Mon Jul 11, 2016 11:00 am

Re: which was the best X11(R4)-terminal?

Unread postby Y888099 » Wed Apr 19, 2017 11:22 pm

miod wrote:VxShitWorks


Why your comment is so negative? VxWorks is a RealTime OS, certified to be used in avionics, automotive, etc.
And what is the alternative? Linux? Windows CE?

User avatar
jan-jaap
Donor
Donor
Posts: 4766
Joined: Thu Jun 17, 2004 11:35 am
Location: Wijchen, The Netherlands
Contact:

Re: which was the best X11(R4)-terminal?

Unread postby jan-jaap » Thu Apr 20, 2017 12:27 am

Y888099 wrote:Why your comment is so negative? VxWorks is a RealTime OS, certified to be used in avionics, automotive, etc.

Having more than a decade of vxWorks experience under my belt, let me say this:
* The licensing costs are very high. If you want vxWorks 653 (certified for critical operations), licensing costs are infernal.

* vxWorks BSPs do not sufficiently abstract the hardware. Every BSP implements certain things you take for granted in a subtly different way. Things like interrupt handling. They generally do not configure the PCI bus, except for hard coded configuration of PCI devices on the SBC.

* As a result of the above, if you support multiple SBCs you end up writing your own OS abstraction layers, and to support 'N' different SBCs running 'M' different versions of vxWorks you end up shipping and supporting MxN different builds which is a nightmare.

* Most people still demanding vxWorks do it because they're entrenched in it. That means even though vxWorks supports kernel/userland separation these days, they demand DKM binaries, which means everything runs inside kernel space and there's zero protection. Heck, everything shares the same namespace: if two processes have a function 'f()', the last one loaded overwrites the first one. Ugh.

* Much of the hardware that runs vxWorks is very specialized. Access to documentation or even a spec sheet can be a problem. I'm not talking about signing an NDA, things like ITAR may prevent disclosure of such information to a non-US resident like myself. A violation of ITAR can cost a US company hundreds of millions so they take it *very* seriously.

* You mention avionics etc, but that's not vxWorks but vxWorks653 which is an entirely different animal. '653' is indeed certified for crtitical operations, and they mostly did that by starting from something arcane (vxWorks 5.4 I believe) and stripping out everything that's useful. It is a total nightmare to code for. And don't stare yourself blind on the 'flight approved' glamour: in reality this means your code has to be audited and certified as well, which is a lengthy and unbelievably expensive process.

The whole vxWorks ecosystem is the opposite of e.g. open source: it's secretive, requires support contracts and money. I can imagine that miod doesn't like it ;)


NB: If you want something running X11R4, try an Indigo running IRIX4 :D
:PI: :Indigo: :Indigo: :Indy: :Indy: :Indy: :Indigo2: :Indigo2: :Indigo2IMP: :Octane: :Octane2: :O2: :O2+: Image :Fuel: :Tezro: :4D70G: :Skywriter: :PWRSeries: :Crimson: :ChallengeL: :Onyx: :O200: :Onyx2: :O3x02L:
To accentuate the special identity of the IRIS 4D/70, Silicon Graphics' designers selected a new color palette. The machine's coating blends dark grey, raspberry and beige colors into a pleasing harmony. (IRIS 4D/70 Superworkstation Technical Report)

User avatar
smj
Donor
Donor
Posts: 1650
Joined: Mon Nov 12, 2007 7:54 pm
Location: Berkeley, CA, USA, NA, Earth, Sol
Contact:

Re: which was the best X11(R4)-terminal?

Unread postby smj » Thu Apr 20, 2017 1:05 am

Y888099 wrote:
smj wrote:NCD-19b, -16c, etc - 68k-based models, before any MIPS or 88k models)


Weren't 68Ks (020? 030? 040?) a bit slow?

Consider how late Apple was selling Macs that ran the same 16MHz 68000 that had to run current MacOS. For the time, it was very useable so long as you didn't need lots of graphics operations. And remember, your choices for X11 GUI were all incredibly plain and basic compared to anything you've seen on Linux in the past 18 years or more.

I reckon prime time for the 68000-based models was 1988-1992? By 1994 they were only for non-technical users at my shop. NCD and others were able to ship faster CPUs as prices came down, or in higher-end, higher-cost models.

If you want to talk slow, a few firms introduced low cost X Terminals trying to use the 8086 or 8018[68]. Ooh, 640x480 monochrome on a 14" screen! It might have been due to design trade-offs, building down to a price, or just plain bad software, but the ones I saw were poor performers when they were still current.
Then? :IRIS3130: ... Now? :O3x02L: :A3504L:- :A3502L: :1600SW:+MLA :Fuel: :Octane2: :Octane: :Indigo2IMP: :Indy: ... Other: DEC :BA213: :BA123: Sun, DG AViiON, NeXT :Cube:

Y888099
Posts: 393
Joined: Mon Jul 11, 2016 11:00 am

Re: which was the best X11(R4)-terminal?

Unread postby Y888099 » Thu Apr 20, 2017 4:27 am

jan-jaap wrote:Having more than a decade of vxWorks experience under my belt


I am an EDA guy. Fellows in the upper class (software engineers) give me directions, other dudes give me mechanical dimensions, and I send back to them schematics and PCBs (sometimes also FEM and other analysis). I have actually never put my hands on their software. Judging the enthusiasm they put on VxWorks, I believed it was a top quality piece of software.

I have some skills with linux, and a few embedded PowerPC boards (supported by linux), so I wonder if tinyX/directfb + libx11/libxcd is a good idea for an home made xterminal.

User avatar
miod
Posts: 477
Joined: Fri Oct 09, 2009 2:44 am
Location: Clermont-Ferrand (France)
Contact:

Re: which was the best X11(R4)-terminal?

Unread postby miod » Thu Apr 20, 2017 11:38 pm

Y888099 wrote:
miod wrote:VxShitWorks


Why your comment is so negative? VxWorks is a RealTime OS, certified to be used in avionics, automotive, etc.
And what is the alternative? Linux? Windows CE?

In addition to what Jan-Jaap said, here are a few annoyances out of many:
  • VxWorks ships with two compilers. The Diab compiler, obtained when VxWorks bought Diab Data a long time ago, and the GCC. Theoretically, you are given the ability to use either compiler. In reality, some parts of the VxWorks source code do not build out of the box with GCC, and some parts do not build out of the box with Diab.
  • Of course, if you are working in avionics, you don't really have a choice, because only Diab is certified, so you have to give up using some parts of VxWorks (this is not as bad as it seems, as these parts are usually not those you'd want to run on a plane).
  • When I say GCC... it's a modified old GCC (IIRC it's a 4.3 nowadays). Linked against their licensing code, of course; and the documentation hints that there have been a few changes made to the compiler, but you don't get the source code. If you want to use a more recent compiler, for example in order to get better code generation for the system you're targetting, then you need to build it yourself. Been there, done that, multiple times, and it's a bumpy road every time.
  • VxWorks claims their environment is POSIX-compatible. But actually, only the userland environment, where you're working in "kernel+N processes" (RTP) mode, is. The kernel part still has many incompatibilities, from the type of interfaces to differences in return values. And since there is too much legacy code, the official communication from VxWorks is: "we will not change any of this, but we will nevertheless still pretend we're POSIX compatible, without any details". Good luck trying to get large chunks of code compilable both in legacy and RTP modes. Even worse if you're trying to keep your code portable between VxWorks kernel mode and other, non-VxWorks POSIX systems. Been there, done that, cursed a lot in the #ifdef wrappers.
  • Windriver has just discovered warm water, yet is annoyingly bragging about it. 6.x brought the "VxBus" abstraction of one IC possibly attached to different busses. Woo hoo! Didn't Linux get something like that in 1994? Not to mention BSD where it is actually a key point of the driver design. Of course, VxBus is only for the new developments. Not all BSP and drivers have been converted.
  • Of course, there is still no decent MII abstraction. Every Ethernet driver handles its own media changes. This is getting a bit better, because they are factoring more and more code in "miiLib" to reduce the occurrences of the media change logic. But then "miiLib" only really supports the original, pre-gigabit, spec; there is some TBI code as well for gigabit phys, but it does not play well with miiLib.
  • Lack of usable dependencies. When building a VxWorks project (either from their provided Eclipse-based IDE, or from the vxprj standalone tool allowing automation), you need to pick one particular application profile, and then add the components you need, either cherry-picking them, or with some preconfigured "bundles". That's fine, so far. Then at some point, you want to shrink your footprint a bit and remove one component you have no real use for. Be sure you have a backup of your project, because their dependency mechanism will gladly unselect as much stuff as possible. Remove the DNS resolver? Fine, then you don't need sockets as all. Oh! Then you have no use for Ethernet at all, let's remove it!
:Indigo:R4000 :Indigo:R4000 :Indigo:R4000 :Indigo2:R4400 :Indigo2IMP:R4400 :Indigo2:R8000 :Indigo2IMP:R10000 :Indy:R4000PC :Indy:R4000SC :Indy:R4600 :Indy:R5000SC :O2:R5000 :O2:RM7000 :Octane:2xR10000 :Octane:R12000 :O200:2xR12000 :O200: - :O200:2x2xR10000 :Fuel:R16000 :O3x0:4xR16000 :A350:
among more than 150 machines : Apollo, Data General, Digital, HP, IBM, MIPS before SGI, Motorola, NeXT, SGI, Solbourne, Sun...

Y888099
Posts: 393
Joined: Mon Jul 11, 2016 11:00 am

Re: which was the best X11(R4)-terminal?

Unread postby Y888099 » Fri Apr 21, 2017 2:48 am

Yup, and Workbench is based on Eclipse, and it sucks like Eclipse. Personally I HATE every Eclipse-based developing systems, but let's talk about alternative: Linux? On GNU ecosystem? When you have to manually configure things and compile them with gcc? Million of reasons to hate it, including how dirty the kernel code itself and its building scripts go. I can't talk about kernels, I simply have some skills, but I definitively don't like it, and it's not my job. So, I don't like linux and its GNU-ecosystem, but it's simply a practical choice on the path of achieving your final purpose, which is not talking about the ideal system as you want to build useful products :D

So ( I assume ) it is with WxWorks: you have to suffer, but at the end you get a good job done. Indeed on Tektronix terminals VxWorks does a good job, check it out: TekxWare/NCBridge is fast, stable, and it takes just 6MB of ram for the whole (kernel + fonts + X11-server, everything in 6MB!!!). WoAh, it's a miracle if compared with the dead elephant linux !!!

edit:
Oh, and about money, it's business which reloads itself

ThinPATH Systems, Inc.
XP400s are fine x-terminals and you can run the last NCBridge.
It's is currently priced at $595 USD

(April 18, 2017)
Last edited by Y888099 on Fri Apr 21, 2017 4:34 am, edited 1 time in total.

User avatar
jan-jaap
Donor
Donor
Posts: 4766
Joined: Thu Jun 17, 2004 11:35 am
Location: Wijchen, The Netherlands
Contact:

Re: which was the best X11(R4)-terminal?

Unread postby jan-jaap » Fri Apr 21, 2017 4:26 am

Y888099 wrote:let's talk about alternative: Linux?

Well, there's always QNX.

Anyway, the software is only part of the picture. I can still get interrupt latencies 3 .. 4 faster on a 10 year old PPC MVME board than on the latest and greatest x64 consumer hardware. These days all buses are serial (PCIe) and there's the APIC which introduces more latency. Nothing beats a copper wire between my device and the CPU :)
:PI: :Indigo: :Indigo: :Indy: :Indy: :Indy: :Indigo2: :Indigo2: :Indigo2IMP: :Octane: :Octane2: :O2: :O2+: Image :Fuel: :Tezro: :4D70G: :Skywriter: :PWRSeries: :Crimson: :ChallengeL: :Onyx: :O200: :Onyx2: :O3x02L:
To accentuate the special identity of the IRIS 4D/70, Silicon Graphics' designers selected a new color palette. The machine's coating blends dark grey, raspberry and beige colors into a pleasing harmony. (IRIS 4D/70 Superworkstation Technical Report)

Y888099
Posts: 393
Joined: Mon Jul 11, 2016 11:00 am

Re: which was the best X11(R4)-terminal?

Unread postby Y888099 » Sat Apr 22, 2017 3:26 pm

Image

robespierre
Posts: 1437
Joined: Mon Sep 12, 2011 2:28 pm
Location: Boston

Re: which was the best X11(R4)-terminal?

Unread postby robespierre » Sat Apr 22, 2017 4:26 pm

InfoWorld wasn't necessarily known for its editorial accuracy. Cyan-Magenta-Blue indeed.
I was also surprised by the X-Window printer concept: complete madness!
:PI: :O2: :Indigo2IMP: :Indigo2IMP:

Y888099
Posts: 393
Joined: Mon Jul 11, 2016 11:00 am

Re: which was the best X11(R4)-terminal?

Unread postby Y888099 » Sun Apr 23, 2017 2:53 am

This article was useful for me to identify the second generation of tek-68k models and the CPU they use: tek25,27,29 / MC68030. Unfortunately I don't have references and notes about their software. Was it Texware? Did Version 7 (released on tek200s) support them? And which was the bootfile? "/tekxp/boot/os.10"? And what about their performances capabilities features and options?

Who knows!?!?

Also, I found a manual yesterday, and it mentions "Frameviewer" in the first page.

Code: Select all

XPCD v7.0 Frameviewer

Unfortunately the second page is missing, so I don't know what is assumed one has to use it for.

And what is Frameviever in first place? Never heard.

Image

Collecting some bits around, I have summarized this table, which compares HP's vs Tek's vs NCD.

Y888099
Posts: 393
Joined: Mon Jul 11, 2016 11:00 am

Re: which was the best X11(R4)-terminal?

Unread postby Y888099 » Tue Apr 25, 2017 2:38 am

So, I have here an HP Envizex P (not mine). It's the fastest version @ 33Mhz, but it's not so nice as expected. It's big, heavy, all ok, but It makes noise. Also, it seems 38MB of ram isn't enough to load a Windows Manager :shock: !?!

Crazy. is the Motif Windows Manager bigger than MIT Windows Manager? Really?

I can configure bin/modules.ld and config/term.cfg to load and execute a Windows Manager instead of trying to invoking the old XDMCP protocol which mandates that the X server starts autonomously and connects to the display manager.

But some weird things have been happening

load vuerx // VUE/RX Workspace Manager ---> not enough memory
load mwm // Motif Window Manager ---> not enough memory
load twm //MIT Window Manager ---> loaded, running

Not enough RAM while there clearly is ... :roll:

I am using TFTP to download the code, I don't want to use NFS instead. But I will try it.


Edit:

Local clients use shared libraries (e.g. libx11, libc, etc). This reduces the amount of memory required for local clients once the first one is loaded. Individually, local clients require the following approximate amounts of memory for NFS:

First xterm: 1.8 MB
Next xterm: 230 KB
First hpterm: 1.6 MB
Next hpterm: 400 KB
vt320: 1.8 MB
Flash File Manager: 2.6 MB
vuerx: 3.8 MB
dtwmlite: 4.5 MB
mwm: 2.8 MB
twm: 1.6 MB
xclock: 360 KB
xlock: 450 KB

With NFS everything works as expected.

N.B.
So, mwm (2.8 MB) is actually bigger than twm(1.6 MB). LOL :lol:

User avatar
smj
Donor
Donor
Posts: 1650
Joined: Mon Nov 12, 2007 7:54 pm
Location: Berkeley, CA, USA, NA, Earth, Sol
Contact:

Re: which was the best X11(R4)-terminal?

Unread postby smj » Wed Apr 26, 2017 1:20 am

Y888099 wrote:load twm //MIT Window Manager ---> loaded, running

I always knew twm as Tom's Window Manager. I don't recall anything being called the MIT Window Manager, but people do all kinds of things I don't know about... :D


Y888099 wrote:So, mwm (2.8 MB) is actually bigger than twm(1.6 MB). LOL :lol:

So those are supposed to be the resident size for a running instance on the Envizex? That used a RISC CPU, right?

I only mention it because the joke was that "mwm" stood for the Megabyte Window Manager, because on 68k or VAX it outrageously took up a whole megabyte or so! This joke was painfully dated though, similar to the expansion of Emacs being "Eight Megabytes And Constantly Swapping" since that originally referred to when 8MB of RAM in a VAX or similar was a lot...
Then? :IRIS3130: ... Now? :O3x02L: :A3504L:- :A3502L: :1600SW:+MLA :Fuel: :Octane2: :Octane: :Indigo2IMP: :Indy: ... Other: DEC :BA213: :BA123: Sun, DG AViiON, NeXT :Cube:

Y888099
Posts: 393
Joined: Mon Jul 11, 2016 11:00 am

Re: which was the best X11(R4)-terminal?

Unread postby Y888099 » Wed Apr 26, 2017 2:34 am

smj wrote:So those are supposed to be the resident size for a running instance on the Envizex? That used a RISC CPU, right?


yup, measured on Envizex's diagnostic mem-usage :)

Y888099
Posts: 393
Joined: Mon Jul 11, 2016 11:00 am

Re: which was the best X11(R4)-terminal?

Unread postby Y888099 » Thu Apr 27, 2017 3:25 pm

A big problem comes if the Xterm doesn't support TrueColor. Both Tektronix and HP are PseudoColor. It seems it's caused by cairo, and every gtk* application is affected.

I don't know if there is a trick to fix it, I know applications like gtkwave, geany, gedit, ... are not properly working, and they are almost useless since you can't understand what you see :roll:

NCD NC400 and NC900 are TrueColor. According to this page Tektronix XP400s are pseudocolor, and so are HP entria and Envivex.

We are doomed :lol:


Return to “Miscellaneous Operating Systems/Hardware”

Who is online

Users browsing this forum: No registered users and 2 guests