[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: I'd rather edit my spreadsheet with a text editor

On Fri, 13 Jun 2003, Andrew Pimlott wrote:

> On Thu, Jun 12, 2003 at 02:27:24AM -0700, Avi Bryant wrote:
> > Never mind that you *could* trivially set up, say, a Squeak image so that
> > you could edit source in a file and run it from the shell - that would be
> > completely missing the point, and you might as well be using Ruby.  This
> > isn't something that would even occur to a non-programmer ("yeah, Excel
> > is nice, but I'd rather edit my spreadsheet with a text editor...")
> I would _so_ rather edit my spreadsheet with a text editor.  I'm not
> an experienced spreadsheet user, but every time I've tried, it's
> become an unmaintainable mess.  How do you debug a spreadsheet?  If
> something's wrong, do you have to inspect every single cell to find
> the problem?  Or is there some better way I'm missing?  (Hom many
> real-world spreadsheets do you suppose have an unnoticed error due
> to a stray keypress?)
> I usually end up rewriting the spreadsheet in a normal programming
> language, with hacked-up spreadsheet-like helper functions, and
> spitting out the results in a crude text form.  At least then I can
> organize things in a logical fashion, include comments, try out
> different variations without getting all mixed up, etc.
> What I would really like is an environment in which I could program
> the logic in a text editor, then map the inputs and outputs to
> grids, graphs, and charts in arbitrary ways.
> Non-programmers achieve impressive results with spreadsheets, and I
> don't begrudge them this tool.  But I suspect there would be
> something better for both programmers and non-programmers if the
> spreadsheet were not so entrenched.
> Andrew

There are several "Emacs spreadsheets".  There are also several systems for
placing in ascii text things that will be evaluated and returned right in
the ascii, which is often html, and the rest is passed on "unevaluated" to
form the whole text object.

Certainly it would be good for a spreadsheet system to make available the
source code for any spreadsheet.  I do not know if any of the popular
spreadsheet systems do this.

The issue is general.  Code is often a collection of objects, such as
source, fasls, relocatable binaries, call graph pictures, test suites,
histories, etc. and we should frankly admit this, and recognize that
category theory is good for something besides names of devices to help
functional programming systems do IO.