Modern math

I DO get it. You have a job and very little math skills. Not surprising.

I entirely agree that a great (too great) proportion of non-trivial conventional software has at defects - but the 99.999% number is made up on the spot, without backing.

In some areas at least, defect rates are going down as people put more effort into quality control. In particular, unit testing can help significantly - you break your code into smaller parts that are well tested, which can dramatically reduce the risk of some kinds of defects. (It is of less value for integration and "long range" defects, and does not help at all against defects in the specifications.)

I can't tell you any real figures here, but I /can/ tell you that the

I am not sure who would win in a competition of the MS product I hate most. I am not a Word fan either. (Again, LibreOffice is better in many ways.)

John S wrote in news:q4ebuj$5s2$ snipped-for-privacy@dont-email.me:

No, you have no clue whatsoever, in fact.

No, you have no clue whatsoever.

You obviously, decidedly do NOT 'get it'. The chrono terminus point is a progressive track. I do not expect a putz like you to be able to grasp that premise either, however.

One step at a time. That is how real men go through life. A dumbfuck like you obviously does not, nor can not qualify as a real man. You are too busy crafting immature, lame, petty jab attempts, like this one was. Grow up, little boy.

ty-Metrics.htm

Was this data processed by a spread sheet or "non-trivial conventional soft ware"?

Isn't a spread sheet program "non-trivial conventional software"?

er

NG

That's a good point. I have often reflected on the nature of test benches in HDL which often rely on the same assumptions of understanding of the req uirements as the code being tested if not the same code itself.

I would often test modules like UARTs against itself which doesn't mean it will work with other systems and in one case it failed to uncover a difficu lt to track down newbie coder bug. lol

Rick C.

h:

yes, WSL is bascially the same as running a linux in a virtual machine

...and the errors are mostly due to those funny 6-legged thinggies found dead between the Excel relay contacts..

And probably not well commented.

And probably not well commented.

Don't blame microsoft unfairly, It's broken that way for compatability with "Lotus-1,2,3"

I prefer gnumeric myself

Of course I blame MS for that. They should not have copied the bug - they should have had it as a per-spreadsheet compatibility option for import from 1, 2, 3. They may not have been the origin of the bug, but they /are/ to blame for their mishandling of it and perpetuating it.

Not well? Not at all!

Exactly. Commenting is a valuable part of procedural programming, or HDL programs, and spreadsheets are seldom (never?) commented. Probably seldom checked.

Ignorant question from someone who doesn't use Excel: how do you control in which order the math is done?

My guys, when they do a simple experiment, type a mess of numbers into a spreadsheet and plot graphs. I just make dots on graph paper with a pencil in real time, as I'm doing the experiment. I win on time.

My graphs have added stuff: my name, date, graph title, part numbers, schematic scribbles, namely context.

John Larkin wrote in news: snipped-for-privacy@4ax.com:

It depends on what you are trying to use the workbook to do. Spreadsheet programs are good at storing and post processing collected data.

Since it is in the form of tables and 'cells' and not in the form of a written program where each step is determined by the program code, it has no way of knowing what you want to happen in a particular order.

If you are talking about simple formula execution rules, then you need to simply be able to analyze each excel function yourself at the time you intend to begin using it to process your data with. If you cannot do that, not only are you not a math peron as you so often claim to be, but you are also lacking in diagnostic skills and other skills needed to evaluate the tools you use.

DUH!

Not knowing order of cell execution makes a spreadsheet useless for anything much more than bookkeeping, and I hate bookkeeping.

John Larkin wrote in news: snipped-for-privacy@4ax.com:

My test sheets were graph plots of the ranges and scales, and the tester placed the points on the graph in real time. Transferring later to the actual spreadsheet to print out and present at the meeting looking very nice and precisely compiled.

I win... on time... acuuracy... presentation, and data retention for post processes. ISO certified.

You... yeah... John... You are 'certifiable'.

John Larkin wrote in news: snipped-for-privacy@4ax.com:

You are a ture idiot. And you would not even rate the skill level of an entry level bookkeeper.

You quite obviously have no clue how a spreadsheet operates, much less how to use one to advance your company.

Transferring data into a spreadsheet and plotting and making PowerPoints or whatever obviously wastes time. My graph is done when the hardware testing is done. I can mark a point on a graph a lot faster than you can type a mess of numbers into a horror of a Microsoft mess. In a thermal experiment, which is inherently slow, my added plot time is zero.

I send my hand-drawn graphs, and whiteboard sketches, to potential customers. They seem to like them. Most everybody is sick of "presentations." Some people can't think without Excel and PowerPoint.

ps - Excel is hardly "Modern math." Visicalc was invented for people who can't program.

I mostly use spreadsheets as scratchpad tools for quick models to test out ideas. They are incredibly handy for creating test datasets or doing a sanity check on what an existing program claims to be doing.

Accounts departments seem to specialise in huge spreadsheets that have no discernable architecture at all beyond "its all in there somewhere".

Short answer is that you don't. Though most do compute from top left to bottom right by default if you force a full recalculation. It has to work no matter which order the computation is run and there should be no circular dependencies at all. Normally when a cell is altered only its dependencies are recomputed and the cascade that results from them.

It means that if you want to iterate to solve some numerical equation you have to tabulate all the intermediate function evaluations and then predict the next value of x to try on the next line down.

You can force manual mode where the updates only occur when you force a recalculation. Handy if you are inputting a lot of data into cells.

Some of the older spreadsheets would permit circular references and run from top left to bottom right. You could program life in those.

It all got a lot more hairy when multiple threads came along and you could sometimes get race conditions between creating a chart and plotting the points on it programmatically (an early bug in XL2007).