[Tutor] pointers or references to variables or sub-sets of variables query.

[David L Neil]

On 8/07/19 10:54 AM, Alan Gauld via Tutor wrote:
> On 07/07/2019 20:54, David L Neil wrote:
> 
>> (However, some of us grew-up at a time when RAM was expensive and even
>> in our relaxed state, such 'costs' still impinge on our consciousness -
> 
> Indeed, my first computer was at the local university and had 64KB.
> My second computer was a Sinclair ZX81 (Timex in the USA?) with 16K

Wow, go you! I could never cope with using analog tape decks as digital 
storage devices.

I've just decided to take a look at SBCs. I guess the 'splash' of news 
about the Raspberry Pi 4 contributed to that, and yet also contributes 
to my prevarication/procrastination...

My 'first' was at high school - an over-grown accounting machine. 
Paper-tape program-software, Teletype-style input and 'line-flow' 
output, and a magnetic drum for data-storage (and yes, I was computing 
physical locations in order to optimise response times - try that!)

At uni, we built Motorola D2 Kits - IIRC an 8-bit Motorola MC6800 
processor. Maybe a 6809 - or more likely, that was the one to which we 
aspired. Those who could 'afford' more hardware courses started building 
intelligent devices, eg embedding a microprocessor within a 'dumb 
terminal'/'green screen'. Intelligent devices, Internet of Things. Plus 
ça change!


> My third, a CP/M machine with 64K and 256K RAM disk and dual
> floppies - such luxury! :-)

Ah nostalgia.

One of my divertissements of that era (early-80s) was MP/M - a bus 
network of what were effectively single-board processors. It implemented 
my first 'computer lab' and we taught everything from COBOL to 
accounting software and word processing on it. Fantastic stuff in its day!
(interestingly, my SBC research last night, took me to a R.Pi device 
embodying exactly these concepts: 
https://www.notebookcheck.net/Raspberry-Pi-Combine-a-Raspberry-Pi-with-up-to-4-Raspberry-Pi-Zeros-for-less-than-US-50-with-the-Cluster-HAT.426998.0.html)


> So I agree, it is hard to get out of that mode of thinking. But
> today the minimum RAM is typically 4GB or more. My desktop
> boxes all have 16GB and even my ancient Netbook has 4G.
> My 20 year old iBook has 640M and even that is enough to
> run Python with many thousands of data objects instantiated.
> 
>> particularly temporary, DB tables into MySQL's MEMORY storage (and with
>> almost zero code-change/risk)!
> 
> Yes, I use SQLite's MEMORY facility reguilarly. Not for managing
> high volumes but where I need flexible search capability.
> A SQL SELECT statement is much more flexible and faster
> than any Python search I could cobble together.
> 
>> (appreciating that I have no difficulty moving from (Python) procedural
>> programming to (SQL) declarative, but many of our colleagues hate such,
>> and with a passion)
> 
> Yes, I've never quite understood why some programmers are
> reluctant to use SQL. For complex structured data it is by far the
> simplest approach and usually very efficient, especially with big
> volumes. But simple searches on small datasets are easier (or as easy)
> in native Python.

Agreed, but if we move beyond standard dict-s, into multi-keyed data 
structures - even with PSL and PyPI at our disposal, isn't it 
much-of-a-muchness to use MySQL/SQLite versus linked-lists or trees?
(or perhaps am showing too much bias from personal experience?)

The "reluctance" (good word!) is intriguing: (a) one more 
package/language to learn - yet such claimants might well have been the 
ones leaping into NoSQL a few years back; and (b) it is a different way 
of thinking - compare 'spaghetti' and monolithic code to "structured", 
procedural to OOP, OOP to 'functional'... I notice a similar likelihood 
to avoid HTML/CSS because of their 'declarative' approaches.

Hey, if you don't like the green ones, that's all the more Smarties/M&Ms 
for me!
(also, avoid the brown ones, they may be 'sheep pellets'/rabbit 
droppings...)
-- 
Regards =dn