Slashdot Mirror
Ask Slashdot: Spreadsheet With Decent Programming Language?
First time accepted submitter slartibartfastatp writes "Spreadsheets are very flexible tools for data analysis and transformations, the obvious options being MS Excel and LibreOffice. However, I found increasingly infuriating to deal with the VBA--dialect functions or (even worse) its translated versions. Is there any spreadsheet that allows usage of a decent programming language in its formulae? I found PySpread intriguing, but still very beta (judging from its latest release version 0.2.3). Perl or even javascript would be better options than =AVERAGE(). Do you know any viable alternatives?"
As old as the hills: http://siag.nu/siag/
Not a very elegant language, but way better than any spreadsheet that I know of.
www.r-project.org/
It's not exactly a spreadsheet, but Pandas is totally awesome and is useful for many tasks for which you might think of using a spreadsheet.
http://pandas.pydata.org/index.html
IPython Notebook is sort of like a combination of the normal ipython shell and an IDE. You interact via your browser but it connects to a normal python process on your local (or remote?) system.
http://ipython.org/ipython-doc/dev/interactive/htmlnotebook.html
I've used these tools together for many tasks for which I might otherwise have used a spreadsheet, particularly for "pivot tables" and time series analysis. Again, even combined they do not a spreadsheet make, but they are in many ways superior. They can handle very large data sets, and best of all you are doing it all in Python.
-73, de n1ywb
www.n1ywb.com
Instead of a spreadsheet with good programming just program and output a spreadsheet. CPAN has plenty of packages for this.
To my knowledge, C# can be used to write plugins for Excel, which should be able to handle the more complex macros.
Self proclaimed typo king, and inventor of the bear destroying coffee table (patent not pending).
https://datanitro.com/index.html
I've used it a bit and it's pretty fantastic
If you want a great spreadsheet: http://www.quantrix.com/
If you want to beef up the programming language but are fine with Excel: http://www.wolfram.com/products/applications/excel_link/
If you are talking non commercial: Siag (suggested above) is cool: http://siag.nu/index.shtml
This hasn't seen much activity in a decade but Haxcel: http://www.johanmalmstrom.se/haxcel/ is Haskell in a spreadsheet.
Yes, both outside COM automation and vSTO plugins that run in Excel http://msdn.microsoft.com/en-us/office/hh128771.aspx
What is the definition of a decent programming language?
It's a pretty ambiguous requirement.
C? Java? Python? Perl? Javascript?
Each is 'decent' in it's own way.
Another way to ask is this: What do you feel the shortcomings of the Excel VB language variant are?
Huh?
Both LibreOffice and OpenOffice.org support macros in Python or Javascript. Chances are you already know one of those, so you don't even need to learn a new language.
Beta is broken and the link to classic doesn't work. Stop wasting our time or there won't be anybody left here.
Wow... quite possibly one option that would be WORSE than VBA.
Self proclaimed typo king, and inventor of the bear destroying coffee table (patent not pending).
"Not all software is versioned with the assumption that 1.0 = finished."
This is the problem. This should be true: version 1.0 should be a statement by the author that the software has reached maturity for its initial, core feature set. Users should be able to rely on this. Version 0 should be when you write your first line of code; less than 1.0 should be initial implementation effort; 1.0 should mean first stable, complete public release; and full-number versions thereafter should indicate compatibility shifts or otherwise large functional jumps.
Let's get together on this, open source nerds. Give up on the fetish for teensy weensy version numbers. I know you think it's cool, but it's not, it makes obvious that you have no regard for the meaning of version numbers. (On the flip side, software like Chrome should stop using full-number increments for every single release. That also razes the meaning of version numbers.)
Stop programming in your fucking spread sheet. It's not an application development system.
If you start having more code than you have data in there, you're doing it wrong.
On the other hand, I got paid a pretty penny to turn a spread sheet system into a real application not so long ago,
I mostly agree, but MySQL is probably the wrong choice of database. For most things that you would consider using a spreadsheet for, you probably aren't concerned with multiple users with concurrent access, so you don't need a DB server, and SQLite is a much better choice. If you do need a DB server for some reason to back your spreadsheet-like analysis, PostgreSQL is probably a much better choice (if nothing else, because of the much richer query functionality; CTEs, particularly, are very useful for analysis.)
What, exactly is wrong with =AVERAGE()?
Excel can already use VBA, which in turn can use IronPython.
Done.
Awesome; but not quite done. At that point you can run an X86 emulator inside it and boot Linux. Then you can run Firefox inside it and finally, you will have access to a sensible language.
Actually, this is one of the best Ask Slashdots ever. A language war enclosed inside a user interface design war enclosed inside a programmer pet hate.
=~ s,(.*),<sarcasm>$1</sarcasm>,g if any_point_you_wish();
What are you doing with a spreadsheet that you find the built-in functionality so limiting?
When the only tool you have is a hammer, every screw looks like a nail. When the only tool you know how to use is a hammer, it is the screw's fault that it won't go in like a nail when you hit it.
I remember many many years ago, the entire corporate sales database where I worked was kept in a text editor. When I pointed out that there was something called "ingres" and that maybe it would be better to keep the data in that, I was told that the person who managed the "database" knew how to use the text editor and didn't want to learn anything else. I was also told that I wasn't hired to write database code for the administration of the company, I was hired to write scientific data processing code.
Part of the problem was that DEC supplied the text editor with the system. That made it the tool of choice, even for problems it was very poor at handling. Likewise, if "large computer company" provides few or no real programming tools other than extensions to a spreadsheet, then the spreadsheet becomes the programming tool of choice. I've had several recent projects that I've had to code in javascript/HTML simply because I couldn't expect them to be used if they were programmed in perl, because web browsers come with the system and perl does not.
"Let's get together on this, open source nerds."
There is already something of an industry-wide standard for version numbers, and it fits with your definition pretty closely. It's just that many don't follow it. Including outfits like Mozilla, in recent years.
The standard calls for major and minor version numbers, followed (optionally) by a build or release number. E.g., 2.3.456.
Version 1.0 is supposed to be the initial, stable, core release, just as you say. But some groups (like Mozilla) insist on jumping the major version for relatively minor reasons, and others seem to get stuck at version 0.5 forever. But that's not because they can't agree on a standard. It's because they just don't follow it.
Spreadsheets are bad at just about everything. Use R instead. If you really need a spreadsheet, there are modules that act like a spreadsheet. But you'll be doing yourself a favor if you wean yourself off the spreadsheet teat.
R is better suited to this type of task than general purpose languages like Python. Most variables and functions in R are vectorised. It's very rare to ever have to write a for loop, which makes the language much more readable.
R is so good at this kind of thing that you don't need anything special to do a pivot table. Just use tapply() and sum(). There's also a 'reshape' package that is far more flexible than anything found in Excel.
Give me Classic Slashdot or give me death!
For symbolic math, Mathematica is vastly superior to Matlab. In my comprehensive exam, Matlab said if I increased the gain of a control system to 1E8, then the following error would be zero. However, for that particular control system, I knew that this result had to be wrong. For modestly large gains, the average of the absolute value of the error should have been a constant, and unaffected by the gain. At gains of 1E8, most physical systems go unstable. The issue shook my confidence on the written portion of my comprehensive exam.
SPICE and Mathematica computed the correct result. The key difference is Mathematica is a symbolic solver. It solves the formulas, without making unnecessary approximations. Spice is absolutely amazing for control system work. It analyzes stuff that most users would be unable to model with Matlab. In particular, SPICE models output to input capacitive coupling correctly, where most other models ignore the issue. Thus, SPICE will frequently predict that a system will be unstable if the gains are sufficiently large, whereas Matlab will often predict everything is good. Additionally, after knowingly blowing the results on the written, I verified the result on a physical system. I wanted to be really sure I had the correct answers for the oral portion of the comprehensive exam.
Matlab is a numeric computation package. In the case of control systems, it quietly converts Laplace transforms into discrete time z-Transforms before computing the system response. Never trust numeric results when they disagree with the theory. To this day, I still wonder if the professor that asked that particular exam question knew about this bug in Matlab, and deliberately asked the exam question from hell.
I think OpenOffice/LibreOffice can be interfaced with a number of programming languages
It can. And even more APIs. Almost all of which are cryptic, cumbersome and/or poorly documented.
It can be worth it once you learn how ... assuming you have enough sanity left.
What, exactly is wrong with =AVERAGE()?
It's not too bad, but it's not too good either.
Yes. I am advocating that version numbers should carry broadly-recognizable meaning. I accept your different opinion but I disagree with it.
In fact I think it would be silly for you to try to defend the suggestion that "version numbers should have no inherent meaning". None? Would you advocate that version numbers be non-sequential? After all, Mac OS 10.6 came after Mac OS 10.5, but maybe next they could release Mac OS 3.6, and then Mac OS 31.5, and then Mac OS -2, and then Mac OS Pi.
Of course version numbers carry meaning. They can't carry lots and lots of meaning, but they can carry a little bit. Why even bother with dotted-decimal version numbers if the dots and decimals mean nothing? Just use integers, but even monotonically increasing integers have "meaning" in that they convey directional advancement of the software. Likewise, if you use dates, dates carry meaning. Other than random numbers, it's difficult to imagine a version numbering scheme that has "no inherent meaning".
And you can do it all in Emacs.
Rule of Slashdot #0: You and people like you are not representative of the larger population. - A.C.
No, Excel won because it was very much better .. at what the majority of users use spreadsheets for: keeping simple lists. Microsoft realized that early on and optimized for it. Excel is also terriffic these days as a simple graph-paper-oriented drawing program: make the cells square and you can outline and color quite easily. It also does a respectable job at turning a imple set of data into a pretty infographic-style graph.
Excels behavior as a tool for complex financial calculations is simple irrelevent for 99% of its users. It won because it was optimized for doing simple, visual stuff.
Socialism: a lie told by totalitarians and believed by fools.
If you can keep it to rslt = function(cell1, cell2, cell3) then it's OK, but in practice it seems to involve rslt = use.of.some.object.you.didnt.expect where goat.sacrifice(was.successful) [but.I.lied.to.you]
Protoplasm. Quiet Protoplasm. I like quiet protoplasm.
What, exactly is wrong with =AVERAGE()?
It's not too bad, but it's not too good either.
Try writing a replacement sometime from scratch, and see how hard it is.
It intelligently only averages cells that are filled with numeric values, allows easy input of multiple ranges of cells, allows direct input of numbers as function parameters, and has an easy to remember name.
If the built-in functions (which include some serious statistics and analysis functions) don't do the job, there are third-party add-ons that likely do. If you absolutely need something unique, then VBA is quite easy to use. The only real thing I don't like about Excel is the "error in a cell is propagated to all cells that reference it", with no way to disable it, and no formatting codes that hide errors. For example, there are a lot of times when I end up with divide by zero because a cell isn't filled in yet, but that's OK (like a table that calculates price/quantity, when a row hasn't been entered yet), and the only way around it is to use the "=IF(ISERROR(...))" construct. It would be much nicer if the existing "positive;negative;zero;text" custom formatting added ";error" to the end.
What are you on about - JS is quite a neat little language for solving small problems, and you can do functional programming in it if that's your thing. There's a reason MIT etc are turning from Scheme (which I love) to JS as a teaching lanuage.
The only part of JavaScript that really sucks is the first four letters, but don't be put off by that.
Socialism: a lie told by totalitarians and believed by fools.
Excel was okay. Word was acceptable.
Wordperfect was excellent, 123 was excellent.
Wordperfect + 123 was twice the price of Word+Excel+Windows, wasn't integrated and couldn't multitask.
Microsoft outcompeted Wordperfect and Lotus by combining the marketplaces so that they couldn't compete. By the time Wordperfect (+Quattro) and Lotus (+Amipro) created their office suites and targetted the Windows platform, it was already too late.
=Woosh() FTFY
Similar concept, but the other end of the technological timeline is the ExtJs grid control (comes with some excellent docco)
You pick up a whole lot of complexity with the ExtJs framework, but you can pretty much implement a whole spreadsheet on it (someone has!), and it's all with Javascript since it's in the browser...
Might not be the same experience as local spreadsheets (no saving to a file :-( ), but it is extensible.
I heard an interesting story from, oddly enough, a MathWorks rep (the people who make MATLAB) about the early days of Microsoft Excel. I don't know if it's true...I think he said he'd either heard it or read it online from an early Excel developer, but my quick Google-fu didn't turn anything up. In any case, he said that after Microsoft first released Excel, they went out to their business customers to figure out how they could improve it. They were flabbergasted to find out that people were using it in completely different ways than they imagined. Even though (I believe) it was originally designed for data analysis, a great number of people weren't even using it for calculations at all. They were using it for to-do list tracking, calendars, structured text documents, presenting tabular data, etc. That's why Microsoft was the first one to have a spreadsheet that allowed the user flexibility to change its appearance: fonts, colors and the like.
He was explaining this as part of his justification for coming out and talking to us, but I think it's also telling that their customers weren't using it like they expected. I guess this is really just a long way of saying that once you get to the point where quick Excel formula isn't cutting it, it stops being the right tool for the job.
Like when I want to actually look at my data in column format (scrolling, frozen panes, column hiding, conditional cell colouring anyone?). Or when I need to edit it (e.g. convert ascii strings to something numeric using search/replace). Or when I want to do a quick interactive pivot table. Or a quick sum or count. Or when I want to try out one or two formulas or expressions before I start coding them. O r when I just need a small table to look good for insertion into a document (the best Latex table editors that I know are plug-ins for Excel or Calc: format in spreadsheet, push button, copy-paste Latex code; works every time).
Of course it's possible to do most of those things in R too, if your time time has no value and if you love writing one-off code. I prefer to select the best tool for the job, and use that. Even if that sometimes means using VBA.
Interestingly I find myself using RExcel (integrating R and Excel) sometimes.
Most of the time however I have no time for zealots who tell me that I don't need X,Y, or Z because I supposedly can make do with A,B, or C too. They can e entertaining though, as long as you recognise them for what they are: rants from zealots.