[Date Index] [Thread Index] [Author Index]
Re: Stephen Wolfram's recent blog
On Friday, February 15, 2013 10:08:24 PM UTC-8, djmpark wrote: > Still it would be nice to know: What is the purpose of releasing a "Core" > > Mathematica language; who is going to use it and what are they going to use > > it for? > > > > My worry is that this is another project dug up from the "to do" list that > > will divert WRI from what, to me at least, seem like much more urgent tasks. > > I thought "a system for doing mathematics by computer" was a nice slogan, > > and I thought Theodore Gray's design of the notebook Front End was the right > > thing. All this led to the impression, which I'm sure WRI meant to give, > > that Mathematica was a medium for writing active, literate technical > > documents. (Rather than being just a graphical calculator or a programming > > input sheet where results are copied out and used elsewhere.) > > > > However, to say that one can write literate technical documents with > > Mathematica is one thing, actually doing it is another. There are design > > aspects of Mathematica that are so awful and difficult that they turn > > Mathematica into a something of a kludge. I would guess that only a very > > small percentage of users use Mathematica as both a writing and calculating > > system - a shame because most people will do better work if they write about > > what they are trying to do, so they could refer to the work sometime later > > when they may have forgotten some of the ideas and objectives, or > > communicate it to someone else to obtain help. > > > > The user interface to Mathematica, in the sense of the types of commands > > available to the user, could be greatly improved. To say that in Mathematica > > "everything is an expression" is a bit like Andrzej's "Anything that is > > spoken by people in England is English by definition." (What about the guy > > who spoke English to his wife, French to his mistress, and German to his > > horse?) It would mean something if "Expressions" were composed of intuitive > > sematic units that could be combined in natural ways. Or if this carried > > through to lower semantic units. This is quite often not the case. > > > > For example in graphics it would be nice if "everything was a graphics > > primitive" and one just drew one thing after another. So axes, frames, and > > various labels might be something you drew instead of using Options. But, > > instead, graphics consists of a large collection of special-purpose plot > > types that are confusing and illogical in methods to combine or modify them. > > Generally a large set of Options is used to modify them or add elements. > > > > Or take the matter of specifying custom tables. Tables are specified by a > > Grid for the content and everything else is specified through Options. > > (Options have their uses but Mathematica is getting to be more and more > > programming by Options. And Option names are getting to be more and more > > Strings. And it's not always easy to get a list of them or what to do. So > > maybe the new language could be called Options?) And the Grid Options seemed > > to be tuned to producing striped or plaid tables. Most custom tables won't > > look like that. Never mind, WRI sort of provided a way and it's up to you to > > figure it out. I wonder how many users design nice custom tables as part of > > their documents? I provided a TableMaker facility in Presentations that > > let's one make a table more like one would draw a graphic. But there is a > > Grid bug such that one can't adjust the size of a row if the row contains > > any spanned elements. That was there in Mathematica 6 and they were aware > > of it, and the bug still there in Mathematica 9. > > > > Or take the matter of writing extended Applications, which more people > > should do. Documentation depends on Workbench and Workbench has never been a > > finished program. There are lots of little inside tricks you have to know, > > and warnings you should ignore. You can't do things like write a Tutorial in > > your own style sheet and have linking done. There is no facility for > > developers to include a template menu. It's not exactly intuitive at all. > > > > All the doo-dads are to me a nuisance. The one thing I liked was the Ctrl+K > > (which I could type faster than you can say Jack Rabbit) command completion. > > But now Ctrl+K is gone and what replaces it is ill-conceived and horrible. > > > > So, please get back to the original concept, polish it and complete it, > > Stephen. > Thanks for the clear words which I think are spot-on. Some may (or will?) call it a rant, and will get defensive about the existing idiosyncrasies, but I think you point to a list of concrete improvements that should be taken very seriously.