windows 1.0 was written in c or pascal?

@@MichaelDavydoff From what is known, it was primarily C and pure assembler. The PASCAL ABI however was used, likely to save space and memory.

@@NCommander is there any reference sources about windowing system programming in c except "Smalltalk-80: The Language and its Implementation"?

@@MichaelDavydoff Are we talking about writing GUI apps or writing a graphical UI system?

@@NCommanderI would like to trace the history of GUI developments over the last 40 years and try to reimplement similar but simpler graphical UI system (windowing system) just for fun and learning

You won't by any chance still have the OS/2 beta SDK stuff or the DDK? That's one of the bigger holy grails in undumped software, especially any of the 1.x DDKs.

Working with Xlib is like carving your UI out of 1s and 0s. Motif on the other hand is using hex :/ I've looked at the Motif programming interface and it just looks painful. Less painful than say Xlib, but painful none the less.

@@NCommander I found Motif OK, it at least had layout managers that would resize automatically rather than the absolute positioning that Microsoft's MFC forced you to use. It was designed with GUI builders in mind though, as it was expected that pretty soon everyone would design their interfaces using a graphical layout tool and not have to code most of the Motif stuff themselves. Interestingly, Microsoft were involved in the aesthetic design of Motif, and their experience with it would influence the appearance of Windows 95.

@@chriswareham If you look at the WIndows 95 Chicago builds, or early Windows NT, the Motif resembance is much clearer. toastytech.com/guis/indexwindows.html has a very good collection of photos to look.

@@NCommander It would be less painful to carve pixels on the screen with Vulcan. I prefer QT on top of anything that deals with the absolute madness that is desktop graphics on Linux or Windows, perhaps with Cairo and glib with Wayland would be another history. But thanks, I pass this challenge. XServer, no, I don't want it. (nor do I want win32/winforms)

I don't use OCRemix music anymore because it has a bad tendency to trip Content-ID even though its OK for commerical use as long as proper credit, but I will say it has a nice effect. I might use it on later/special videos, especially remixes of some of FF7'siconic tunes, or Terra's Theme/World Map FF6.

MFC is still used today. That directly dates back to 1991 ish. Which is a layer over the old C APIs/SDKs. In VS 2019...docs.microsoft.com/en-us/cpp/mfc/reference/creating-an-mfc-application?view=msvc-160

Windows might be slow to create Window or new Processes, but creating threads is super-fast. Linux on the other way is the inverse. I get see the problem with the API being cumbersome to use. Does it work ? so what matters. Its the same thing as asking Linux to drop the stupidity of its process creation execve/fork. Its all POSIX and it works and is documented, so what would be gained replacing just the API ? not much.

I know this comment is old, but I wanted to just say that, yes while the backwards compatibility is impressive, and those APIs are still in use today, they really shouldn’t be. I feel this stuff sort of holds back tech and innovation in a way, similar to Apple’s M1 silicon kind of highlighting how even x86 could be considered outdated. I still think this stuff is useful, and can 100% see how it is still used today, but I can also see the other side where it does hold back innovation and pushing forward the boundaries of tech. Sincerely, A zoomer linux user :)

Not quite as true for WPF. It's a ground up presentation framework.

It is, however, built on DirectDraw and DirectWrite, which you could say are legacy API's

Thank you kindly :), I'm just lucky I recorded my mistakes.

You and MichaelMJD should get together and do a super-retro video :)

@@Reziac I'd be open to some sorta of collobration, my first "real" KZfaq anything was actually guest starring in @RetroManCave talking about NeXTstep. That being said, I might not be big enough, Michael MJD is one of the larger retro tech channels and I only have two videos and 67 subs.

Once upon a time I recommended a fellow with SIX subs to a largish channel of my acquaintance. They had him on as a guest and his channel took off like a rocket. Anyway, I don't think size matters here; this is a specialty interest and this video certainly brought new stuff to the table! Besides, one need only say the magic words... "Soylent News" :D

@@Reziac Feel free to drop Michael MJD or any other channel on my behalf; I'd be happy to experiment with doing something like bridging Windows NT 4 Domains over a hodgepodge equipment of the Internet. I'm actually getting a video edited on NetWare and then going into netbooting DOS and OS/2 from it. I'm not sure I'm comfortable doing so, it feels a little to close to self-promotion if I do it directly. The only reason I felt OK even posting this to SN was that I had the technical writeup to go with it.

Four if you count i386 and x86-64 as seperate archs, since they dropped i386 support in Catalina. But Apple often don't provide long-term compatibility even on the same arch.

even so, Mac Os always had a very good retro-compatibility, even cross-architecture

They used to. Then iOS happened :/. I did a demo showing a bunch of System 1 apps running in Classic on Mac OS 10.4 right before intel macs got announced.

@@NCommander before they removed 32 bit support, certain software like Basilisk and Sheepshaver just ran much better than it's windows and Linux versions

Interesting perspective. I see the history of Windows and Mac a little differently. Apple has been able to change platforms from 68K to PPC to x86 to x64 to ARM with hardly anyone noticing that anything changed. Meanwhile, the transition from x86 to x64 on Windows was _brutal_ and took all of XP and Vista’s lifetime to go mainstream.

I had trouble with PCem when I tried to film this video originally, but I eventually got it properly working and I used it for footage for both IBM OS/2 1.0 on a livestream and an upcoming video for NT OS/2.

I made a comment on cardfile too, without seeing your comments.. It really did do the job very well.. I think that I still had it in Win95, With a bit of Jiggery Pokery & bending ears at MS lol !

This is technically true, but even by that standard, this program is extremely long. A equivalent Toolbox or Motif program is quite a bit shorter, and it doesn't help that the code demo does quite a few unnecessary things, plus 16-bit WIndows required a lot of boiler plate to deal with segment and resource registeration. If I were to re-implement this on modern WIndows, it would be a fair bit shortly, possibly even half in size while still demostrating the same core event loop and dialog creation aspects.

I don't use OCRemix music anymore because it has an annoying tendency to cause Content-ID/Copyright strikes even though OCRemix music can be monetized (specifically allowed by ocremix). That's why SLS Part 2 has no music after three attempts of scoring it all got content striked :/

I believe that's the book that talks about HELLO.C being a scandal at 115 lines of code. My annoted version is 400+, but that's because I wrote a primer to real mode. Segments dies an inglorious death.

​@@NCommander Thanks for replying! Yeah, that's the one. Well, I guess “Hello World” for the Windows is somewhat of a shock to everybody the first time around … at least until one realizes that things like general setup and window procedures follow quite predictable code patterns. Anyway, since yesterday I've also found time to read through your write-up on Medium. I didn't know about the need for MakeProcInstance() in earlier versions of Windows, for example … really interesting stuff. Seeing all this talk about segments and NEAR and FAR pointers and calls reminded me however, why I've never looked back after leaving DOS/16-bit environments for good at the end of the millennium. In that sense: thank you for fighting the good fight for us-tirelessly wrestling with these old operating system and development environments, just so that we can experience the joy of bringing software marvels like Word 1.1a for Windows to life once more…

Thank you kindly. There's also a video up on Windows for Workgroups networking, and sometime this week, I'm going to have a video on Novell NetWare going up. The NetWare video is proven to be difficult to edit but its (slowly) getting there.

It does. DOS 3.3 isn't that different from DOS 5/6 in terms of core functionality.

Microsoft was *very* big on Hungarian notion through this era, and hilariously, when they migrated to 32-bit, they had the 16-bit sizes in the variable name with 32-bit bitwith. Probably why hungarian notion fell out of favo.

The amount shit hasn't changed is pretty impressive all things considered.

Doesn't count, since that's not native backwards compatibility. By that same logic, I could say Linux is compatible with Windows 1.0.

I'm using MSC 4, but confirm you can compile a bog standard printf() Hello World under DOS first, and make sure to carefully follow the install instructions in the MSC 4.0 install manual.

Short answer: sorta, but not really. Windows 1.0 will start on DOS 4-6, but the DOS shell part will hang if launched. THere's also likely other pain points. Best practice is to use Compaq DOS 3.31 or such which support greater than 32 MiB partitions while still being compatible. In practice, the DOS internal API is mostly unchanged from DOS 3.30 -> DOS 6.22, with the only additions really being the undocumented Windows APIs, and what was bundled in.

PROMPT $p$g is actually the default on all non-IBM versions. you can see that in the DOS 2 source code Microsoft posted, it was specifically customized to C> for IBM.

@@NCommander Really? Thought it was only the default starting with around 5 or so. Interesting

@@Lofote The prompt was originally $n$g, in DOS 1.0, and 1.1 but you could not set it. When DOS 2.0 came out The prompt was $n$g ( drive and the greater symbol ), ( yes, I still with a degree in math need to visualize a fish ), but you could change it to $p$g, ( path and greater symbol ). There was a side effect in command.com for DOS 2.0, in that if you only used the default $n$g, it would not allocate the environment variable, ( as shown by the set command ), but if you changed it, it would allocate that memory ( 13b). SELECT was used to install/upgrade refresh DOS without formatting, If you blew up the system files, you could just type the SYS command, but you needed the first two directory entries to be the system files. It was so rare that drive A was hard coded, and I ran most of my installs from my second hard disk.

Using edlin? That's exceptionally brave :) If I seriously wanted to program on Windows 1.0, I'd probably try to get the DOS MPW workshop to talk to MSC 4.0. If there's any interest, I might do a follow up where I either do that, or see if I can get this to work through something like a DOSbox emulator for seamless development from a modern host system. I know OWCC uses the former method to handle compiling their help files because the source was missing for the help compiler.

I'll def. check it out. This video isn't anywhere as polished as it should be, but it was my first attempt at doing a tech video instead of gaming, but I still consider it watchable even though I've come a long way since then. Maybe 9 years from now (when this channel is 10 years old), I'll redo this :)

I liked it just fine. As my advisor said, when I was lamenting some perceived misstep during a past project, "You can always invent new things."

There was also an OS/2 version of CodeView. You could use it to debug 16-bit protected mode OS/2 apps. It still works in ArcaOS 5.0.

@@zoomosis There is also a Xenix version of CodeView which appears to be able to work on DOS applications through VP/IX.

@@zoomosis What a great pre-IDE tool!

The best way to describe it is that it's re-arranging blocks of memory. It's sorta like taking a toy house and shaking it very vigerously. If there are programming bugs relating to memory access, it should cause the program in question to come crashing down. Unlike later Windows, any single application could crash the entire system.

@@NCommander but why would you expect rearranging a program's memory to not cause problems on correct programs? Is it rearranging objects that are meant to be accessed through Windows handles instead of raw pointers?

@@scose you got it in one; you're not supposed to manipulate segments directly under Windows nor are you supposed to use pointers, you get a handle and use that. In practice, Windows returns a pointer, so that bit of advice meant that real mode Windows programs (1.x/2.x) go *boom* on Windows 3+. Global memory can move unless locked (GMEM_LOCKED, which is a NOP now). In practice, Windows tends to leave things where they lie, so unless you did something that would force memory to move, it wouldn't actually crash. Windows performs segment fixups on the fly, and for Win286/386/3.x it also performs virtual mapping through segmentation fault handling. The New Executable and OMF format specifically encode FIXUPP information to allow segment offsets to be fixed as a O(1) operation. It's also why Dynamic Link Libraries (DLL) are position dependent on Windows and they have a lot of oddities compared to other platforms.

My actual boxed copy of PC-DOS 3.3 basically writes that out in full :)

CP/M has no relationship to UNIX, and PC versions of GEM can't multitask because Apple sued Digital Research. The Atari TOS/GEM versions can (or well, MultiGEM can), but neither are related to UNIX. You might be thinking of DR-DOS (which was based on CP/M-86) and had some multitasking capabilities, but those weren't based on any AT&T code.

@@NCommander Good mornign NC ... _I think I am right. At the time Unix was, "the universities OS which in practice very little people on really understook the power of what they had in hands". Unix is the root of Linux and it was the root of CP/M. GAry Kildall picked up a number of students from university and the team developed the CP/M there ( so far I understand ). They "stripped Unix" and created an OS for PC´s because the processor then, the Zilog Z80 our the Intel 4004 where not powerfull enough to handle much dumb terminals, but so far I understand it was "Unix". Tim Peterson made a copy follwoing the same root and later sold the QDDos to MSFT. I need to find the link where this is explained. I used CP/M, GEM and CP/M 86 in the seventies. CP/M 86 was presented. I am not a programer and I am just a businessman that used PC´s for managing my company since late seventies, early eighties. I coordinated the team that developed what is today called ERP/HR/CRM in Cobol and that I start using in my company in something like 1980/81 or so. Brazil for example, contrary to the world trend then, "closed borders and focused in Unix and Universities" developing their OS´s and aplications on Multi-tasking and multi-user enviroment. This gave Brasil a significant power in their software industry. Am I wrong ? Pls re-confirm because this is history now but I need to update my "museum"

MSFT was obliged to re-write the entire MD-DOS after they got the IBM agreement. Digital Reseach did not sue them for copywrites because they knew they would face IBM lawers and MSFT had the time to re-write MS-Dos. Contrary to what many people opinion, I think that MSFT delayed the computer industry many years. Around early nineties a big chunk of MSFT sales were on the "systems division". Such systems division was "based on Novell" which were able to "re-do MS-DOS" and transform it into a multi.user multitasking OS. All this time was lost because "all the world bought the MSFT single user/single task" concept not realizing their mistake.

@@NCommander www.ipopisp.com/marksofesteem18

www.ipopisp.com/marksofesteem18

For Hello, and maybe FontTest, possibly. At a minimal, you'd have to convert to C89 format since it's pre-standard C (aka old school function prototypes), remove the segment format, convert the resource format from Windows 1.0 to 3.1 to 95 (Borland Resource Converter can do this), and probably fix a few prototypes. Cardfile has handwritten assembly so that would have to be manually converted to 32 or 64-bit. Honestly, I might try this and make a video out of it.

From memory there were datatype changes for many function calls with the move to Win32. Windows programming books written at the time (1993-1998ish) had chapters on what to change. Some of them written for Visual Basic and Delphi programmers. Incidentally it's largely forgotten now but Visual Basic 1.0 was a bit of a "killer app" for Windows development, especially shareware.

I'm not entirely sure if the BGM is something I'll keep on later videos. I've kept it for my videos up to this point.

@@NCommander haha. it is fine, just that I usually listen to FF songs, so I got confused.

@@ngjerome5490 Well, you know, It's Difficult to Stand on Two Feet when you're Under the Rotting Pizza while dealing with Still More Fighting, when you're expecting a One Wing Angel to bring in the Staff Roll. (ok, that last one was forced :) )

@@NCommander subscribed.

It's not the *worst* one I've seen, and there are things that could be simplified about it. Motif is probably the worst I remember if you do anything beyond an about box.