The Vent and Rant Thread

[Freeshadow]

Oh yes an old nerd thread (someplace I could enjoy being a youngster )
Dennis, nicely summed up dos history for those who came after us, although I admit I had no idea about unices back then and that there was a unixoid DOS shell replacement is completely new to me.
I was 12 when I came to computing (386-SX 16; 1 MB RAM; 40 MB HDD; VGA and a 24 dot matrix printer)
I began with DR-DOS 5.0 (OS/2 thereafter), so my first shock was after I had to switch to MS-DOS 5 was to discover there was no xdel (or similar) command... MS' deltree came with 6.

[Freeshadow]

Originally Posted by DMcCunney

No, it's not an adult. It should be, but adulthood is an outgrowth of maturity, and some folks never attain it despite chronological age.
______
Dennis

^that.
IMO said client behaved extremely rude because the attitude summed up was in a nutshell "I can't be bothered with my part of the contract, so pre-chew it for me."

[DMcCunney]

Originally Posted by cromag

I still have one! Well, it was a 7300, but I slipped in a half-height 40MB hard drive as soon as they were affordable (by the standards of the day). So my "wedge" doesn't have a monitor pedestal. The "user agent" had a weird Y2K bug -- it went from 1999 to 19100 -- but the core UNIX was okay.

Yeah, it had the Y2K bug, which won't get fixed. But IIRC, you can reset the date/time after booting to the current one and it will behave.

The user agent was a Convergent creation called FACE. You could plug an ANSI terminal into it and get a usable mono character mode display similar to what you saw on the console. I thought it was one of the better UIs I'd seen. (I used Daniel Lawrence's version of MicroEmacs as my editor under MSDOS, and Lawrence's source built out of the box on the 3B1. I had fun customizing it to do something appropriate when I pressed one of the dedicated keys on the console keyboard. I had even more fun doing the same thing for an older version of Gnu Emacs.)

A client of the employer I worked for back when had a 3B1 with 2MB RAM, a 72MB HD, a serial ports card, and had four Wyse terminals and a printer attached, running a specialized distribution management application based on a UNIX RDBMS. Performance was acceptable.

There was an amusing bit early on, when the client ran up a large LD phone bill. The machine had been shipped to the app vendor's office in Boston to get the DBMS installed, and they had it set to poll their local server for updates. They forgot to turn off polling when they shipped it to NYC for installation, and it matter of factly dialed out over the modem daily to check their server, because it had a dial-up number for the server it wanted to poll. Oops... The vendor agreed it was their fail and split the charges with the customer.

(I contributed a fair bit to getting it working, and got offered a job by the vendor. I declined because I didn't want to leave NYC, but have wondered since if I should have accepted.)

I used to turn it on for the kids to show them what a "green screen" looked like.

I used to have an amber Amdek monitor plugged into my PC clone, driven by a Hercules graphics card. That was considered a European DIN standard because it was believed to be more readable and less likely to cause problems like eyestrain in use.
______
Dennis

[DMcCunney]

Originally Posted by Freeshadow

Oh yes an old nerd thread (someplace I could enjoy being a youngster )

I am officially an Old Phart these days, though I'd like to think I maintain a youthful outlook.

Dennis, nicely summed up dos history for those who came after us, although I admit I had no idea about unices back then and that there was a unixoid DOS shell replacement is completely new to me.

There were actually several, including DOS implementations of the Bourne and C shells. I collected an assortment. The MKS Toolkit simply offered the most complete and polished solution for the PC. Aside from the Korn shell, there was a full version of the vi editor and a few other things. I was able to use the DOS PRINT command with some Korn shell aliases to craft a decent approximation of the Unix lp print spooler.

(MKS no longer exists as such, but the Toolkit became the basis for Interix, which underlay Microsoft Services for Unix package.)

I was 12 when I came to computing (386-SX 16; 1 MB RAM; 40 MB HDD; VGA and a 24 dot matrix printer)

I was over twice that.

I began with DR-DOS 5.0 (OS/2 thereafter), so my first shock was after I had to switch to MS-DOS 5 was to discover there was no xdel (or similar) command... MS' deltree came with 6.

The first computer I dealt with was an IBM mainframe compatible at a bank I've mentioned elsewhere. It was the days when the original IBM-PC running DOS 2.1 was appearing on corporate desktops as an engine to run the Lotus 1,2,3 spreadsheet. I watched bank officers write memos as Lotus text fields because they knew how to use it but didn't know how to use a word processor or have one installed on their machine.

IIRC, DR-DOS originated from requests by DR customers for a version of DOS that could be embedded in ROM. MSDOS couldn't back then - to be ROMmable, the code and data segments needed to be separate, and weren't in MSDOS. When it got to a reasonable state of development, DR decided to release it as a commercial product and alternative to MSDOS, and it gained a reasonable following.
______
Dennis

[DMcCunney]

Originally Posted by PeterT

We almost need an I'm an old nerd thread.

My first exposure to computing was in Grade 8 and was via a 2741 terminal to IBM using APL. The next year we switched to IP Sharp and their APL environment. Skip forward a few years and we also had an IBM 5150 running APL. The next year (my final High School) we moved to an HP 9830A which used Basic, and had both a cassette tape for storage and a card reader (pencil marked, not punched).

I never had to deal with APL. An old friend is an APL specialist, and once worked in the APL group at IBM. He recounted talking to an APL vendor on behalf of a client, and saying "I need to see your source code to confirm it is suitable for my client."

"You can't do that. We don't provide source."

"Okay, I'll disassemble a binary."

"If you can do that, we'll give a you a doctorate!"

"I already have one."

"Okay, we'll give you an office next to (APL creator) Ken Iverson!"

"I used to have one."

Game, set, and match.

He still programs in APL, using a descendant called K, and claims to be able to deliver completed applications in the time other developers spend figuring out how long it will take them to do it.

After that it was off to University where I was exposed to Burroughs 6700 via timesharing, an IBM 360 via punched cards and LSI-11s running RT 11. Of course in first year my friends and I used the LSI 11s to enter our assignments, sent them to the 6700 and on to a high speed card punch, before feeding the card deck into the card reader on the 360. It sure made getting nicely formatted punched cards easy.

We had card readers at the bank, but I don't recall them ever being used.

I dealt with virtual punch cards, as 80 column card images in a PDS file. Much more convenient than paper, and no need to play "52,000 pickup" when you dropped a card deck.

Then that summer off to a job where we had an Amdahl V6 and a Dec-10, and a conversion project was underway from MFT to MVS. I still marvel at the fact that the V6 had a whopping 8 MB of memory, and it supported most of the data processing for an international company!

Never dealt with Amdahl kit, but mainframes could do amazing things.

When I arrived, the bank I worked at had built out a regional data center, and managed to bring it in for just under $1 million. (I saw the original RFP.) To do so, they went third-party and plug compatible. The mainframe was an ITEL box that was 370 equivalent, and supported 500 remote 3270 terminals through TCAM and CICS under OS-VS1. It had 2MB physical memory and 16MB virtual memory. It was fine, when it worked, but outages were frequent, in part due to not running current IBM OS releases because the machine wasn't from IBM.

I had a photocopy of a cartoon from Datamation in my cube, with the field engineer walking into a customer site saying "System been down long?", and addressing the question to a skeleton covered with cobwebs in a swivel chair in a cube set up like mine. Someone made a copy of it and taped it up on the inside of the VP of IT's office door. He was not amused.

He had also come up through the ranks as a COBOL programmer, and there was still a module on the system he maintained to keep his hand in. He sat down at the terminal to do some work on his program, and Lo! The system was down. The systems programmer spluttered and turned purple when I told him about it. He dreaded on line demos, for good reason.

They subsequently replaced the ITEL box with a pair of genuine IBM 4341s running MVS, loosely coupled under JES2, and reliability soared. Of course, around the time the data center truly hit its stride, the decision was made to re-centralize everything back at the Division level. (Never mind that the two year applications backlog at Division was what made the region build out it's own data center...)

Those were the days!

They were indeed. I enjoyed it, but I'm not certain I'd want to do it again There are too many new and different mistakes to make.
______
Dennis

[CRussel]

Originally Posted by DMcCunney

(MKS no longer exists as such, but the Toolkit became the basis for Interix, which underlay Microsoft Services for Unix package.)
______
Dennis

Well, not strictly true. The developers of Interix were former MKS folks who wanted a "purer" solution, but they didn't use any MKS code that I'm aware of.

To set qualifications for this comment, I should state that I ran Interix (called OpenNT at that point) in it's first semi-public implementation, and again when it officially released. I was later the first (and only) Microsoft MVP for Services for Unix (SFU), and was involved as their contract tech writer from the very beginning, writing the initial whitepapers for SFU v1 (which was based on a very limited subset of the MKS Toolkit). MS acquired Softway Systems shortly after the release of Interix 2.2 and just about the time of Services for UNIX 2.0, which did NOT include any Interix code, but was purely based on the MKS code they had licensed.

It wasn't until SFU v3.0 was released that the Interix subsystem was included in the product, completely replacing the MKS code. With SFUv3 and later SFUv3.5 (the final release), the Interix subsystem completely replaced the (less than satisfactory) NT POSIX subsystem, giving Windows Server a fully POSIX compliant subsystem that ran as an equal partner to the Windows subsystem.

ETA: I should also add that some of the people acquired when MS bought out Softway Systems ended up being key in the development of PowerShell, for which I am still a Microsoft MVP.

[DMcCunney]

Originally Posted by CRussel

Well, not strictly true. The developers of Interix were former MKS folks who wanted a "purer" solution, but they didn't use any MKS code that I'm aware of.

Thanks for the clarification. MSDOS code would present challenges in porting to an NT environment.

<...> (various fascinating history elided)

I ran Services for Unix at one point, and also ran Cygwin and AT&T's UWIN package that Dr. David Korn developed. These days I run a set of the Gnu utilities built "native" for Win32 with MinGW. I never encountered the early OpenNT flavor.

ETA: I should also add that some of the people acquired when MS bought out Softway Systems ended up being key in the development of PowerShell, for which I am still a Microsoft MVP.

I've looked at PowerShell, but haven't done a lot with it so far. I'm intrigued by the notions of manipulating fully qualified .NET objects in scripts, but thus far haven't needed to do it.
______
Dennis

[Katsunami]

Originally Posted by DMcCunney

He still programs in APL, using a descendant called K, and claims to be able to deliver completed applications in the time other developers spend figuring out how long it will take them to do it.

Shiiit.... I've looked it up, and that language is actually worse than regular expressions.

I can believe that guy, however. I've more than once seen people trying to do stuff using if-statements for things such as validation, replacement, and such. With some work, it is often faster and even easier to use regular expressions, and with a lot less code as well.

For example, I have an embedded web application, which actually runs on top of PHP. As it has no database, it uses a file called "app.conf.php". The settings are like this:

define("SETTING", "value");

Instead of reading the file line by line into an array of strings, finding the setting, replace the value and then writing back the entire file (which is the way I've seen such things done most often), I just use a command line call to 'sed', and replace the value using a regular expression. (And yes, a take into account that someone could have modified the file, and the setting is now stored as define ("setting" , "VaLuE") )

Thus, I effectively replace (almost) an entire function with one line of code.

I can imagine that APL can do the same.

[DMcCunney]

Originally Posted by Katsunami

Shiiit.... I've looked it up, and that language is actually worse than regular expressions.

APL is intended to perform matrix operations. The guy in question writes financial applications, and was writing code to support bond traders at Morgan Stanley at one point. APL is just made for stuff like that.

A nice bit about K is that it dispenses with the special APL character set used to describe the operations it supports. (The old joke in the "Shoot yourself in the foot" classification of programming languages is in APL you shoot yourself in the foot, then spend the next couple of days figuring out how to do it in fewer characters. )

I can believe that guy, however. I've more than once seen people trying to do stuff using if-statements for things such as validation, replacement, and such. With some work, it is often faster and even easier to use regular expressions, and with a lot less code as well.

Regular expressions are wild cards on steroids and are extraordinary powerful, but can easily become complex enough the developer ceases to understand them. I've seen at least one app intended to parse and validate REs and help you understand what they are doing and confirm that they are doing what you want.

For example, I have an embedded web application, which actually runs on top of PHP. As it has no database, it uses a file called "app.conf.php". The settings are like this:

define("SETTING", "value");

Instead of reading the file line by line into an array of strings, finding the setting, replace the value and then writing back the entire file (which is the way I've seen such things done most often), I just use a command line call to 'sed', and replace the value using a regular expression. (And yes, a take into account that someone could have modified the file, and the setting is now stored as define ("setting" , "VaLuE") )

Thus, I effectively replace (almost) an entire function with one line of code.

Which is what you want, as long as you (and anyone else who has to look at it) understand that one line of code.

I can imagine that APL can do the same.

Likely.

For another take that is arguably more powerful than regular expressions, see SNOBOL4 patterns.
______
Dennis

[Katsunami]

Originally Posted by DMcCunney

For another take that is arguably more powerful than regular expressions, see SNOBOL4 patterns.

I must not read things such as these. They make me wish I was born in 1960. Then I would have lived through all the good computer stuff, and I would now be much closer to retirement.

I caught the last whiff of it, because I started on a very old XT in 1990, running DOS 3.2, and later 5.0, with only two 720 KB disk drives. So, I worked like it was te 70's or 80's until 1994. Then I was thrust into modern computing with OS/2 3.0, and later Windows NT.

That love of the old, hardcore CS stuff never went away though.

My education was Computer Science, with a specialisation in (very low level) programming of micro controllers because of the love of the hardcore stuff, but nowadays, I'm often writing HTML/CSS, and trying to understand the latest Javascript framework / library fad.

(I have just written an entire rant about that some pages back... the one that got this entire thing started )

Yesterday and today, I've spend over a day to get a fracking user interface right, to show it at least somewhat similar in the major browsers (IE 11, Firefox and Chrome on the desktop, Safari on iOS, and Chrome/Android Browser and Firefox on Android).

A friend of mine, who is a front-end developer, could have probably done this in less than one day, and add jQuery transitions, slides, and other shizzle-wizzle to make it really smooth and blitzy. Or he would have used Bootstrap or something, which costs too much time to get into right now.

I can't. I'm just happy when the UI looks logical, somewhat decent, and works without problems.

Then, the data ends up at the server, in the backend and that's where I live. I write the stuff nobody ever sees.

I juggle data as if it's made of sushi, eating a piece every so often (hey... I do that with the rolls that fail, you know... so I eat invalid data as well). Analyzing it, transforming it, doing whatever the frack you want with it, saving it on disk and/or database... no problem. Done in a jiffy.

That same friend who can make the blitzy user interfaces was a colleague of mine some time ago, and we have 'helped' each other to meet deadlines.

Let's just say he doesn't want my help on UI's because they look like a work straight out of the 80's. I haven't the knowledge to make them glitzy and modern; I have to go look up everything. So while I *could* do it, it would be very time consuming.

Similarly, I don't want his help on the backend/data part... although if "crashing" was a requirement, I would. Crashing the application is the main perk of his backend code. He forgets half of the code, or just doesn't have the knowledge on how to do it right; just as I don't have it for doing nice UI's fast.

He writes front-ends, and does that very well.
I write back-ends and/or algorithms, and I do that very well.

Trying to swap our tasks results in either very ugly, or very crashy programs.

The pity is, the user interface is the stuff people see. Therefore that guy has a nice portfolio that he can show to people, while I can only show pictures of systems I've written code for. Nothing to see here... it's all inside. If you *SEE* my code, you'll probably looking at some sort of crash dump.

I would probably have loved being in the computer science field in the 70's and 80's (the time without screens and proper keyboards might have been a bit too old even for me).

[DMcCunney]

Originally Posted by HomeInMyShoes

I've lived that forever. I remember when C++ first came out and all the job postings were looking for people with 5+ years experience in it. Yep. There's not a lot of people available who actually created the language.

Tech job requirements are ludicrous a lot of the time.

A friend decades ago was reading a job ad, and the employer wanted a PhD physicist with post graduate experience in magnetohydrodynamics and a few other things, and commented "This is a raid. There are perhaps 5 companies in the country doing work in that area, and the guy they want works for a competitor and they want to hire him away!" There's a lot of that.

A chap I knew got hired years back by a competitor of his former employer, and his new manager asked him for the source code for the product he had worked on at the previous employer, and was nonplussed when he stated he hadn't brought it with him when he left the previous job, and wouldn't provide it if he had. New manager had no concept of professional ethics and thought that sort of theft was standard behavior. Maybe it was where she had worked before...

When I interview people for my company I look for enthusiasm and what they think is cool about what they have previously done. I can ask specific questions about technology, but I'm not overly certain if you're just good with google and have a good memory or not.

I am and do, and understand completely.

At a former employer, I had fun, because we were acquired and merged, and part of the process was coming up with the new org chart and titles and job descriptions. I finally got an official piece of paper from HR informing me I was part of a group reporting to a Regional VP of IT. That was all very well, but while the position existed, the person who would fill it had not been selected. I wonder whether I was supposed to write my own performance reviews and sign off on my own timesheets in the interim.

In a rare case of putting a round peg into a round hole, the job went to the co-worker I thought should get it. My boss thought he wouldn't because he didn't have the full set of knowledge about all the technologies in use they wanted. I said nobody had that, and he didn't need to. His responsibility was to know enough to make sure he had people who had the needed expertise working for him, and to understand enough to be able to manage them and evaluate them come performance review time.

I survived through several rounds of M&A related layoffs before I finally got cut too. I had a grimly amusing exit interview with HR where I was given a list of ages and titles of other IT folks who got cut, and it was a clear case, reading between the lines, of "We aren't discriminating against you because you are an older employee - look at the age range being let go. We are laying you off because you are making more money than we want to pay." (Of course, HR was expanding while the rest of the company was shedding headcount and salaries...)

At lot of the IT job listings I see are exercises in wishful thinking on the part of the prospective employer's HR department. I look at the requirements, and say "Good luck getting half of that."

(And I got mortally tired of hearing from Indian high tech recruiters who couldn't grasp why I was unwilling to relocate halfway across the country for 6 month consulting assignment. )
______
Dennis

[wodin]

Never forget, HR is NOT your friend.

[DMcCunney]

Originally Posted by wodin

Never forget, HR is NOT your friend.

I got disabused of the notion they were decades ago.

I had an interview years back where I clicked with the hiring manager, and by the time we finished we were already talking about precisely what he planned to have me do when I came on board. Then HR decided I was over qualified and said no. In a rationally managed company, the correct response would have been "HR, the hiring manager has decided who he wants. Your job is to process the paperwork. You don't get a vote in who he selects." Alas, they weren't rationally managed. The hiring manager was unhappy, but HR did get the deciding vote.
______
Dennis

[cromag]

Originally Posted by DMcCunney

Yeah, it had the Y2K bug, which won't get fixed. But IIRC, you can reset the date/time after booting to the current one and it will behave.
...

Yep, I did that. But by then the Amiga 3000 was my primary computer.

Originally Posted by DMcCunney

The user agent was a Convergent creation called FACE. You could plug an ANSI terminal into it and get a usable mono character mode display similar to what you saw on the console. I thought it was one of the better UIs I'd seen.
...

I did a lot of writing on it. I found the Alpha/graphic Windowing system very useful when I was at the keyboard, but I turned it off when I was on a terminal -- it took a long time to draw and re-draw. I did a lot of writing in the field on an Epson Geneva PX-8. Plugged it into an RS-232 and uploaded quickly when I was using the command line.

[DMcCunney]

Originally Posted by cromag

Yep, I did that. But by then the Amiga 3000 was my primary computer.

I had little exposure to the Amiga. What I knew of it was interesting, but I had no call to deal with one.

I did a lot of writing on it. I found the Alpha/graphic Windowing system very useful when I was at the keyboard, but I turned it off when I was on a terminal -- it took a long time to draw and re-draw. I did a lot of writing in the field on an Epson Geneva PX-8. Plugged it into an RS-232 and uploaded quickly when I was using the command line.

Screen draws were reasonable on Wyse terminals emulating VT100s connected via a cable at 19,200 baud and the character mode FACE UI was usable. But I spent time hacking TERMCAP (the Unix terminal database) back in the day, doing things like trying to work around the "Magic Cookie glitch", on terminals where screen attributes took up a space, and could generally make things work. I had code in my .profile that looked at the TTY I was coming in on and set the $TERM variable appropriately.

I did do sysadmin chores from the command line, and largely still do. Under Windows, I use an open source application called ConEmu which is a tabbed console app. It's possible to have multiple command lines open in tabs in the ConEmu window, and use several different shells. I have CMD.EXE, PowerShell, TCC-LE (a freeware lite character mode version of JP Software's Take Command GUI command interpreter, a successor to the old 4DOS COMMAND.COM replacement under MSDOS), and Win23 ports of Bash, tcsh, and zsh. I can have all of them open at once in ConEmu tabs, along with character mode apps. Fun, for suitable values of the term.
______
Dennis