Programming

[Anonymouse]

I've used DOS 6.22 and 3.1 - did the job, slowly.
95 - ditto, quicker.
98 - slightly better.
2000 - I liked it.
XP - a bit flashy, but OK.
Vista - mmm. Brief acquaintance (long story I'm not going to tell). I admit to liking the Aero interface - my laptop never had speed issues.
8.1 - loved it.
10 - don't like it at all.
11 - like it about the same, i.e. not at all.
I've also used a BBC Micro, a TeleType (stop laughing, this was 1982!), plus SUN Workstation, UNIX and VAX/VMS terminals, programming in UNIX, various BASIC dialects, C (a joy - using Jackson Structured Programming tends to reduce a program design to the ideal of all being functions), COBOL (a pain), PASCAL (ditto but worse, pointers are so confusing), Modula-2 (the bas...illegitimate child of PASCAL - avoid like the plague!), C++ (2 weeks is not long enough to learn it - hell, two years wouldn't be enough!), PROLOG (5, I think - hated it), Visual BASIC 5 (I quite liked it, inventing a technique - or so I thought - that turned out to be SOP), VB.NET and DCL (v. powerful if you know what you're doing). Also 6502, 80386 (not bad, actually - HiSoft Assembler was very capable), Z80 and 68000 Assembler (loved the last two).
Cautionary note: on my HND/degree course I used the VT220 and VT320 keyboards. The < and > symbols were on the same key, needing SHIFT for the latter. I missed this once. One typo, just ONE, i.e. a misplaced < that should've been >, and it cost me EIGHT BLOODY HOURS of debugging a COBOL program! Just to make it even more baffling, a fellow student had virtually the same program, with the same design (JSP tends to force this, especially in COBOL - one advantage of COBOL with JSP is that it's difficult to misinterpret) - but his program worked perfectly with his and my data file!
I swear, I was just about to lose the will to live when, after EIGHT BLOODY HOURS, I finally spotted it. An excellent example (so my lecturer teased me later) of a logical error as opposed to a compiler error - the compiler, of course, never caught it as it was a perfectly legitimate program statement; it only counted as a typo because it was the wrong symbol, not an illegitimate one. Never was GIGO so true!

Oh, I miss the VIC-20/SHARP MZ700/ATARI ST, 6502/Z80/68000 days. Back then IT made sense.

[spiderplant]

Originally Posted by Anonymouse

One typo, just ONE, i.e. a misplaced < that should've been >, and it cost me EIGHT BLOODY HOURS of debugging a COBOL program!

Should have used "IS GREATER THAN"

[Anonymouse]

Can you do that in COBOL? I know it uses (well, used) more keywords than BASIC, but I'd no idea you could say such a thing. Mind you, it makes sense - it did tend to use words rather than symbols, for clarity. It was intended for, mainly, managers, i.e. non-technical people. Come to think of it, I think you could say PLUS and MINUS instead of/as well as + and -. You could definitely say ADD 1 TO COUNT as opposed to COUNT = COUNT + 1 (or, in C, COUNT++).

If it's so, I wish I'd known it then! You have no idea how STUPID I felt debugging that damn program! I even modified the design at one point, thinking I'd found a mistake - not realising the mistake was mine. It should've been obvious: my friend's program and mine had exactly the same program actions, the same design - so in essence they were the same program.
(But this wasn't plagiarism - JSP tends to force this in COBOL. Program actions in the JSP design translate directly to program statements, and so the same design produces the same program. The advantage of JSP is that if things go wrong you can quickly figure out why.
Unless, of course, the error is subtle, as mine was...)

So their giving different results - especially since his program worked exactly as predicted with both data sets, i.e. his program + my data = the results I'd expected for my program, and his program + his data = the results he'd expected - should've made it obvious where the problem was (we were required to work out what the results should be, then check them against the actual results, as part of our documentation - Pat Diskin, our lecturer, was red-hot on that. But you couldn't get away with just going down the list of results and idly ticking them off, oh no - she was shrewd enough to be able to tell, and she would mark you down. No, you had to genuinely check).

It was just that the results were totally unlike what I'd expected, so I couldn't figure it out from the pattern of errors (in fact, as I recall the results were totally screwy and made no sense whatsoever). But I made that ONE correction, and that put everything right. I did confess to the error in my programming log.

If anything, this was useful to teach the importance of thoroughness in checking, accuracy and, above all, GIGO. The resulting statement was logically correct, so the compiler didn't, couldn't catch it. It was just factually wrong - not a syntax error (which would have been caught), but a logical one. No compiler could catch that, even today (well, maybe an AI compiler, if there is such a thing - and even then it'd be required to analyse program context and think 'ah, he's said this but he meant to say that', likely a monumental task even for the Enterprise-D computer).

To coin a phrase, whoops.

I did get an A, though. The corrected program worked perfectly - my program + my data = my expected results, and my program + his data = his expected results. Yes, he got an A too, and Pat did not for one instant think either of us had copied from the other - she could tell that, too. You know, McGonagall reminds me of her...

Originally Posted by Anonymouse

Can you do that in COBOL?

Yes you can, you can also use NOT (as in 'IS NOT GREATER THAN').
Of course, thats the same as 'IS LESS THAN OR EQUAL TO' which you could also use.

Originally Posted by Anonymouse

Can you do that in COBOL?
I know it uses (well, used) more keywords than BASIC

Used ?

COBOL is still alive and well.
I used it reqularly at Experian, and its still used by many mainframes.
Before COBOL, I programmed in FORTRAN, thats still alive and kicking as well.

Dont be fooled by Universities trying to tell you everything is done in so called "Modern" languages.

Can you do that in COBOL? I know it uses (well, used) more keywords than BASIC, but I'd no idea you could say such a thing. Mind you, it makes sense - it did tend to use words rather than symbols, for clarity. It was intended for, mainly, managers, i.e. non-technical people.

No, it wasn’t - it was intended for COBOL Programmers*, who in the 70s and 80s wrote most of the World’s business Systems in it (and you’d be amazed how many Legacy Systems are still running on it).

I was a coder in the 80’s and early 90s, using Mainframe Assembler, COBOL, RPG2, and SQL - never saw a Manager writing code (although they would sometimes give advice, if they were ex-coders).

[Anonymouse]

Using LESS THAN OR EQUAL TO - I'd forgotten. Yes, I remember now, I could have done that. Should have - stating IS GREATER THAN would've been completely unambiguous and I couldn't have got it wrong - or if I somehow had, I'd have seen it straight away.

1989. 36+ years ago. I still can't believe that some of you young 'uns - ooh, shades of Rimmer, I cannot believe I just said that! - weren't even around back then.
I miss those days.

Originally Posted by Hugh

No, it wasn’t - it was intended for COBOL Programmers*, who in the 70s and 80s wrote most of the World’s business Systems in it (and you’d be amazed how many Legacy Systems are still running on it).

I was a coder in the 80’s and early 90s, using Mainframe Assembler, COBOL, RPG2, and SQL - never saw a Manager writing code (although they would sometimes give advice, if they were ex-coders).

Oops, my bad. I should've said business-oriented users. And no, I wouldn't be amazed, or even surprised; I studied programming, data processing and business applications at HND level. The powers that be really want to get rid of COBOL as an outmoded language, and admittedly the new ones are much better, but the simple truth is that without some drastic and unimaginable upgrade, to somehow do it all at once, they can't. It's too deeply embedded, with billions of lines of code, in society-critical applications such as the stock market, which has to keep running 24/7/365 even when trading is closed for the day - because, of course, there are business users all over the world in multiple time zones (even, if the country's big enough, in the same country, e.g. the States). Business runs 24/7/365, so these COBOL programs have to as well.
Unless they create a parallel system and gradually change over to it...which they won't be able to afford to do. Not until we mine Psyche, at any rate.
Which, in a classic Catch-22, we won't be able to afford to do until we get the IT systems sorted out.

"Is Orr crazy?"
"He sure is."
"Can you ground him?"
"I sure can. But first he has to ask me to. That's part of the rule."
"Then why doesn't he ask you to?"
"Because he's crazy, He has to be crazy to keep flying combat missions after all the close calls he's had. Sure, I can ground Orr. But first he has to ask me to."
"That's all he has to do to get grounded? Just ask?"
"That's it."
"Then you can ground him?"
"No. Then I cannot ground him."
"AAAAARRRGGGHHH!"

- Catch-22, the excellent movie

No, everything is not done in the modern languages, that's quite true. They just wish it was. Bit like Brian Arthur, who confidently declared he was going to bring economics into the 20th Century, only for his old mentor to ask sadly, "Don't you think you should bring it into the 19th Century first?"

Hell, I once had a program crash on me at a workplace - and I immediately said to my supervisor, "That's what you get for relying on COBOL."
"Eh?"
"That's what it's written in."
"How do you know?!"
Answer: the error message I got started with %COB-F-, that's why, i.e. a fatal COBOL error. On the VAX, error messages were - still are? - in the format:

%PROGRAM-SEVERITY-ABBREVIATION followed by an explanation, e.g.:

%COB-F-COMPERR, COBOL compiler error - check program syntax

or

%DCL-W-ACCESSREST, access restricted - consult your system operator

SEVERITY was F, S, W or I (Fatal, Severe, Warning, Informational). Can't remember what the error message was, but I know a COBOL core dump when I see it. <ironic>I saw too many of them in my 2 years studying COBOL on my HND, especially the 2nd year. </ironic> Trust me, it does not lend itself to what's known as flat code. I hated that in year 2. That plus FMS (Forms Management System) on the VAX...uuurrggghhhh.
I never used ORACLE on my HND course; that came to the VAX in 1991 or thereabouts, and I was converting my HND to a B.Sc. Whenever two or more ORACLE users were logged on, my PASCAL compilation time went from blink-and-you'll-miss-it to several minutes. In a race between a VAX running ORACLE and a continent, I was sure the continent would win. You wouldn't believe the I/O and CPU time ORACLE used.
Is this still true today? <sarcasm>Wouldn't be surprised.</sarcasm>

I studied ways of modernising systems, such as parallel running...and the usual method of shoehorning a new system into where the old one was. They cannot do that on a large scale; the task is too large, and the programs too critical. Some systems are so critical that people could die if the new ones don't work...which they never do. Hence they must keep going.

Imagine if Tron was true. Picture it: two old programs, as old as Dumont if not older, taking a break for a few hundred micro-cycles (i.e. overnight, after trading has closed for the day):

"Hey, you know the Users can't do without us?"
"Yeah. Pete-One says they want to replace me with some new flashy program, but they can't. I'm old, even outmoded, but still critical."
"Me, too. I've been running nonstop for trillions of micro-cycles - I think the Users call it 'years', or something - but they still depend on me. Got at least a million man-years - is that right, Tron? (a nod) - man-years invested in us."
"All hail the Users, though."
"Yeah. All hail the Users."



"Look, you know how it is. You just keep doin' what you're told to, even if it seems crazy, an' you hope like hell your User knows what's goin' on. Right?"
"Well, that's how it is for programs, yes, but -"
"I hate to break it to you, pal, but most of the time that's how it is for Users, too."
"Stranger and stranger...!"

- Tron, a movie which really made you think

[VX1953]

Originally Posted by djmagnifique

I really liked Windows ME. So much more stable than Win 98.

I liked WinME also, when I was an NHS IT Techie back in the very early 2000's we did a trial batch of Win98se PC's upgraded to WinMe and it worked well for general Windows usage but had problems running certain legacy Novell and custom Dos based programs so it never got rolled out, we stuck with Win98se until XP arrived and that was rolled out Trust wide.

Home wise, I still have 2 Desktop PC's with Win7 on and prefer using Win7 to Win10 on dual boot, Linux is good.

[heero_yuy]

Rather deviating from topic but just about everything round you are
programmed in embedded "C" (My specialty) bear in mind Linux ( and derivatives) like your mobile are also using this.

[Anonymouse]

Oh, C. I did love that, at Kalamazoo in Manchester, as part of some City & Guilds course or other, running on UNIX, and again in 1998 on NT (3.5 or 4? Can't remember.)

I did like the '98 lecturer. I remember one time a student switched off his PC without shutting it down properly (he thought he was doing the guy, who'd stepped out for a bit, a favour); this tended to screw up NT PCs, especially on the multi-user system we had. The next day the student was, luckily for him, off sick. The lecturer (I think he was from Saudi Arabia) said, "I wait till he's better, then I kill him!" .

At Kalamazoo we used JSP, which forces you, at base level, to base all your program modules on functions - which is the C ideal anyway.
I pointed this out to our lecturer. "The only way you can reconcile program actions with C is to reduce everything to functions."
"That's the idea. A C program should be functions anyway." (He was a man after my own heart, describing the difference between hacking and cracking. The former is something I've done, i.e. getting into the guts of the system, finding out how things work and making them work better and/or differently*, whereas cracking is breaking in illegally - he rightly pointed out that 'hacking' was a sorely-abused term. This was in '94, but it's still true today. A hacker is not a cracker, and vice versa.)
"Hmm. Makes sense."

The thing is, C made sense. This is one reason why UNIX/Linux is written in it. Pointers were easy, and they too made sense. I actually wrote a hex monitor program in C, to decode a .EXE file into ASCII, so you could read any embedded text messages. A line would look like:

0AF0 41 20 70 72 6F 67 72 61 6D 20 74 6F 20 61 73 63 A program to asc

with nonprinting characters represented as dots. I later migrated it to Win 98 and 2000 (I dual-booted them), and it worked fine. Watney used a similar hex editor in the film version of The Martian to hack the Rover so it could talk to Pathfinder.

You just try that in BASIC!
(Can be done, but it's much more difficult.)

* A good example of this was on the SHARP MZ-700, which used a built-in cassette drive. Loading up a program from the Machine Code Monitor (the MZ-700 didn't have, e.g., BASIC on ROM, so it could run any version of BASIC, FORTH - ooh, I forgot about that! - or any other language) caused it to automatically run, which was no good if you wanted to create a backup. Wasn't as simple as copying the tape - I know, I tried. But you needed to know the start address, the execution address (not necessarily the same) and the file length in order to copy a file via the MCM. Simple enough; S (Save) <start address> <end address> <execution address> filename to save a block of memory, addresses in hexadecimal.
So after reading the monitor manual (so helpful!) to figure out how the loading process worked, I wrote a Z80 Assembler program (all praise to HiSoft, whose Devpac Assembler program was brilliant!) to intercept a program while it was loading and stop it auto-running, delivering said data. I successfully backed up BASIC and a favourite game, a low-res (characters only) Defender clone.
This had to be done - we (oldies, anyway) all know how vulnerable cassette tapes were, and I was doing a course in BASIC; losing that tape would've crippled me. This was backing up, not ripping off (though it later occurred to me, guiltily, that it could be used for that purpose, oops ).
I wrote to the user group, detailing my hack - and I was awarded a £25 voucher! Result!
I miss that computer. Superb keyboard (I consider myself a keyboard connoisseur - VIC-20 excellent, MZ-700 same, ATARI ST good, VT220 and VT320 on the VAX excellent despite the < > issue mentioned earlier, first PC...eew). Flexible design because it was a "clean" computer, no OS or BASIC in ROM. Built-in and very powerful & useful MCM, FULLY DOCUMENTED. They don't make 'em like that any more.

[nomadking]

Try searching for what would know have been called a memory leak.
Using FORTRAN IV, no debuggers around. Adding a trace statement simply moved where the fault appeared. Weeks upon weeks searching, only to find out it was a complier problem.


In those days if you needed a database, you had to do itself. If you needed to do various transformations of a model in 3D space, then again you had to do it yourself using A level maths. No getting a Microsoft library to handle it all.


Computers and programming have changed so much in the past 50 years.

[Anonymouse]

dBase III? I used that on my HND, had fun with it. Good software. Creating forms was a doddle.

I hesitate to call ORACLE and SQL a database. 'Bloody nightmare' is closer. The version on UNIX c/o Liverpool University, where I did a post-grad course in 1997, was so slow as to be unusable, despite our using SUN workstations which definitely were not slow. Taking screen dumps (required for the course, to show your program in action) was almost impossible and took ages. I despised it. Structured Query Language (I called it unStructured Queer Language, much more accurate - with apologies to our gay patrons) was anything but consistent. A trigger that would work on one data field refused to work on another of the same type. I even remember one which worked one day and didn't want to know the next day - and each module was only 2 weeks long.

Certainly not enough for C++, which I found totally confusing. 2 years wouldn't have been enough IMO. Haven't bothered with it since. Same with Java. My program compiled cleanly, and that was as far as I would (could) go.

Definitely enough for wrapper CGI programs to combine C and HTML generated via C, though. I had fun creating an online kitty adoption shop (some students went overboard with elaborate credit card coding, but I stuck to KISS and made it COD - simples!). Got an A.

Things have indeed changed. Mostly they've gone worse.