Data Entry Efficiency
System
Posts: 45
This discussion was created from comments split from: fastspin compiler for P2: Assembly, Spin, BASIC, and C in one compiler.
Comments
Back in the 90s i helped a friend convert his Qbasic accounting package to windoze with VB3. One customer refused to go the windoze path because it couldn’t keep up with her data entry. She was totally correct tho. And now everything is connected to the cloud, data entry is frightfully slow compared to the 70s and 80s. Thats progress.. its intuitive but horridly slow.
Agreed, the Amiga was a lot ahead, I had two A2000 (I still have them).
Regarding data entry I imagine you are referring to the slowness of moving focus between fields with mouse and then typing again the data. But VB3 already had the tab_index property and events like gotfocus and lostfocus that, together with object's naming property and support for arrays of objects, were capable of building nice GUIs with good automation so that a given form didn't require any mouse interaction and data entry was possible only through keyboard (even using function keys and shortcuts).
I didn't feel any increased slowness compared to a dos program or even an AS/36 AS400 mainframe of that time with their green on black terminals.
Younger operators never knew the speed of the old text based programs so they just think the slowness is normal.
By comparison, in the late 70s, the mini I worked on (taught both hardware maintenance and programming courses for Singer/ICL) which had a memory cycle time of 2.2us and many clocks per instruction (but high level instructions) supported 35 video terminals doing live online data order entry.
With text based VB forms and its controls I never had problems. But I also faced the slowness of graphical parts and when I wrote the HMI for the our vacuum metalization machinery (evaporation, sputering ...) at the time I do everything in VB3 and for the graphics drawing I put an empty picture control on the form, pass its window handler to the dll library where I wrote all the graphic tracing functions that was then called from VB3 in C++ (with VC2.0)
As per compilation, IIRC, VB5 added classes to basic while compilation belongs to VB6 (after this release the VisualStudio saga started). Its nice to see how many people of different ages visits this forum, I am always fashinated, as in the 70s(2) I was born
That is why it won't die.
Mike
I’ve been referring to online, not batch.
COBOL was great for batch, and wouldn’t have been bad for online on the Singer/ICL mini... cobol translated to about 2.2:1 assembler instructions, but we were hamstrung with memory limitations (and its cost).
That mini was built from 1970-1993 and maintained until end of 1999 (and beyond in some places), with only one substantial update redesign) in 1981. M&S (Marks & Spence)had them in all stores in the UK, and in BBC (now Bunnings) - hardware stores - in Oz and NZ.
IN Oz, CSR had one hooked to over 120 remote terminals live at a time via 9600 baud leased lines and Scitec stat mixes. I designed and built the boards that plugged into the mini’s bus to support the remote terminals.
pics 4 posts below
-Phil
-Phil
And no, I don't trust the cloud either! Wait till they start really jacking the prices up and hold you whole operation to ransom.
As for data entry efficiency I can only say that when I did POS terminals that they had to supported very fast keyboard scanning and n-key rollover since operators could be extremely fast, especially before scanning really took off. One of my keyboard tests was to run my finger down a section of the keyboard as fast as I could but I could never beat it. This was originally all on a humble 2MHz 6502 based system running Forth and handling printing and scanning etc under interrupts while maintaining tasks for communications etc.
Yep, LibreOffice is a no-brainer but I've still not managed to convince many to use it.
First pic is my NL90 board that plugged into the ICL System 25 minicomputer's bus (IIRC up to 12 of these pcbs could be added).
The pcb has a Xilinx XC2018-70 FPGA, a pair of Z86C9112's, up to 8 Z8530's and up to 16 RS232 serial ports each with an RJ45 connector.
I hand routed the FPGA to achieve the circuit I required, including delays. Still have the layout printout somewhere. You cannot so that any more!
The second pic is an S25 with the front cover removed. Those RJ45 flat cables (on the right side) each connect to a remote terminal via Scitek Statistical Multiplexers.
The third pic is a [pair of S25's, with a HDD (65MB removable over a 130MB fixed) in the middle. Remember the 'washing machine' sized HDD's anyone
Yup! Worked on 300MB, 11-platter Ampex drives in the mid-80s.
Then came the 11 platters with 40MB. When doing maintenance on the heads a locking pin was used to prevent retraction in the event of a power fail. Otherwise, look no fingers
The heads were aligned using a custom CE disc pack worth around $25K. This was specially written data tracks which was an oscillating as in wavy. Ie if you think of a track as being a straight line then this track(s) would be like a repeating sine wave. With the CRO setup displaying lisagous figures, you would adjust the heads, one by one, to get a proper cats eye display - both sides equal. Cats eye being like an infinity sign. Maintenance was required every 3 months to ensure the heads did not move too much, otherwise you’d never be able to read the old data.
There was other adjustments made also, such as adjusting the head delatching mechanism for power fail. If this wasn’t adjusted properly then a power fail would result in a head crash! Goodbye data and lots of $$$.
I used to teach this stuff. Nothing like getting a brand new (customers) drive and break it down into pieces for the class to put back together
Postedit
IIRC cartridges started appearing around 75-76 and were around 5MB. Thing were moving quite rapidly then and a lot of new companies were coming onto the scene with small minis.