Modern math

I am at the Excel workbook stage.

Reply to
DecadentLinuxUserNumeroUno
Loading thread data ...

In my last job - at Haffmans's BV in Venlo - the computer manager wouldn't dream of letting anybody put anything but Windows on the computers he managed.

I wanted to do some number crunching, and suggested putting a Linux partition on my computer, downloading gcc so that I could write a Fortran program to do the work - an old-fashioned approach, but one that had worked for me in the past.

The computer manager regarded this as a dangerous heresy and a totally unacceptable security risk, so I did my number crunching in Excel - which was horribly tedious, but could be made to work.

I suspect that a sufficiently ingenious hacker could do bad things even if confined to Excel, but the computer manager didn't have that much imagination.

--
Bill Sloman, Sydney
Reply to
bill.sloman

In the UK, the Ordnance Survey offer Excel functions to convert between their OSGB grid locations, on a transverse Mercator grid, to/from WGS84 latitude and longitude. I used them to further create KML placemark files for display via Google Earth...

Mike.

Reply to
Mike Coon
:

t dream of letting anybody put anything but Windows on the computers he man aged.

tion on my computer, downloading gcc so that I could write a Fortran progra m to do the work - an old-fashioned approach, but one that had worked for m e in the past.

you don't need linux to run gcc

Reply to
Lasse Langwadt Christensen

There isn't a GCC version that runs on Windows, surely??

--
This message may be freely reproduced without limit or charge only via  
the Usenet protocol. Reproduction in whole or part through other  
 Click to see the full signature
Reply to
Cursitor Doom

cygwin or mingw

Reply to
Lasse Langwadt Christensen

Google spreadsheet errors

One estimate is that 90% of spreadsheets have at least one error.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

In what way is that surprising?

Reply to
John S

+1.

Cygwin if you want most unmodified *nix programs to compile and work with minimal or no changes, Mingw if you want the fastest environment with less *nix emulation, possibly more program changes needed.

Clifford Heath.

Reply to
Clifford Heath

Cursitor Doom wrote in news:q4c11e$gj2$4@dont- email.me:

mingw gcc

Reply to
DecadentLinuxUserNumeroUno

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

Sure... data input and operator errors. Not so much processing errors.

And 90%? You really are gullible, eh?

Reply to
DecadentLinuxUserNumeroUno

John S wrote in news:q4c5ve$gp$ snipped-for-privacy@dont-email.me:

For not getting it, you truly are an abject idiot.

Reply to
DecadentLinuxUserNumeroUno
:

on win10 you can do even better with Windows Subsystem for Linux

Reply to
Lasse Langwadt Christensen

But that estimate may be one of them.

--
Bill Sloman, Sydney
Reply to
bill.sloman

rg:

n't dream of letting anybody put anything but Windows on the computers he m anaged.

tition on my computer, downloading gcc so that I could write a Fortran prog ram to do the work - an old-fashioned approach, but one that had worked for me in the past.

But I did need a less anxious computer manager before I could download anyt hing and put it on my PC at work. The charm of Linux was that I had it on m y home computer, and knew where to get what I'd need.

--
Bill Sloman, Sydney
Reply to
bill.sloman

Something like 30% of large codebases on "big name" software projects by Google, Adobe, etc. have "copypasta errors" in code of the form like this, a shit-headed way to compare two arrays for reverse equality:

if (array_1[9] == array_2[0] && array_1[8] == array_2[1] && array_1[7] == array_2[2] && array_1[6] == array_2[3] && array_1[5] == array_2[4] && array_1[4] == array_2[5] && etc...

(I made an error I had to go back and fix just making this small example of the problem. Don't write this kind of code...)

Reply to
bitrex

Be careful with your naming here - mingw-w64 is significantly different from the original mingw, and a far better choice. Your link is to the correct version, but the names can be confusing.

Reply to
David Brown

Make sure it is mingw-w64, not the old mingw project. That way you get modern tools and proper C support, instead of one that uses MS's slow and limited C library DLL.

I haven't tried this myself (since I avoid Win10 as much as possible).

A key difference between this and the other solutions is that WSL attempts to run native Linux executables directly, rather than running programs compiled for Windows. That could be an advantage in some cases, but completely misses the point when you want to use gcc to compiler programs for Windows.

Cygwin is a fairly solid POSIX emulation layer for Windows. It emulates things that Windows doesn't support (or doesn't support well), like "fork" and differences in the file system. That makes it slow and cumbersome, requires extra DLL's, it feels "alien", and software has to be under the GPL - including anything you write yourself and compile with its tools - unless you pay for a license. Cygwin has a huge range of packages - as well as gcc it includes *nix utilities, X, gui software, etc.

Mingw (the original) is a 32-bit gcc port that uses MS's C runtine DLL's. It is closely tied with the msys project that provides Windows ports of a range of command line tools, libraries, etc. Mingw programs integrate much better with Windows, and are significantly more efficient than Cygwin programs. But it is getting a bit old, and the use of MS's C libraries makes programs slow and limited (the C libraries are mostly C90 only).

Mingw-w64 is a fork and update that can make 32-bit and 64-bit Windows binaries, and uses modern, statically linked C libraries. It is tied with the msys2 project.

If you want to use a proper, modern C (and C++, Fortran, Ada, etc.) compiler for Windows, you want mingw-w64 gcc. Don't bother with the old mingw project, and don't bother with Cygwin (if you need that level of POSIX compatibility, use a POSIX system, not Windows).

Reply to
David Brown

Although that might be true most of the errors in common spreadsheets are of no consequence such as a random bit of text in an unused cell.

The great thing about spreadsheets is that they have entirely different systematic errors to a procedural computer language so that if you use a spreadsheet to generate test data it is highly unlikely that you will make a serious mistake on indexing.

That is favourable when compared to the 99.999% of non-trivial conventional software that contains at least one defect.

I have seen some truly horrific spreadsheets that have grown like topsy over the years and become utterly labyrinthine tangles of spaghetti. Accounts departments tend to harbour the worst offending stuff.

You can do some very sophisticated data analysis in Excel if you don't mind waiting a while. The only thing to watch is that some of the more esoteric special functions are themselves flawed as is the random number generator which is definitely *NOT* what the Excel documentation claims it is (and at one point produced negative numbers in the range 0-1).

formatting link

This is probably the worst bug in Excel and has been there almost forever - their random number implementation is incorrect. That is the code is not an implementation of Wichmann-Hill as claimed since 2003.

formatting link

It was still wrong in Excel 2013. I haven't tried any more recent versions. It isn't a good idea to do Monte-Carlo simulations with an RNG of unknown provenance and unproven behaviour.

Random algorithms do not random numbers make.

--
Regards, 
Martin Brown
Reply to
Martin Brown

63.6% of all statistics are made up on the spot with no evidence. That applies to both of these.

Agreed.

But I am not convinced that it is the worse bug in Excel - there are so many to choose from. A classic one is the fictional leap year in 1900. While that does not affect many people, MS's bullying, bribery and blackmail process of getting ISO certification for their absurd "OOXML" formats meant they forced through this fabrication as an ISO standard for bug-for-bug compatibility.

For most (but not all) purposes, I find LibreOffice Calc outclasses Excel easily - it is simply more logical and less quirky.

Reply to
David Brown

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.