[Date Index] [Thread Index] [Author Index]
Re: precedence for ReplaceAll?
On 6/30/10 at 1:49 AM, siegman at stanford.edu (AES) wrote: >In article <i0cjk9$88h$1 at smc.vnet.net>, >Bill Rowe <readnews at sbcglobal.net> wrote: >>I am not sure what outcome you are looking for here. >Something like the following: >* Wolfram identifies, from user test panels that they set up, or >from monitoring particularly frequent queries or complaints in user >forums like this one, those "gotchas" in Mathematica that are >particularly likely to be encountered by novice or unsophisticated >or unsuspecting users, or particularly likely to cause serious >damage. Given postings from Daniel Lichtblau, John Fultz and others at Wolfram, I am sure this happens at least on an informal basis. >("Particularly frequent" might be defined as the top 1% of all such >encounters.) >* And Wolfram then attempts to forestall these damaging encounters, >or attempts to assist users in recovering from them as quickly as >possible, by, as appropriate: >--Adding brief warnings about those specific gotchas in prominent >locations in the ref/ or elementary tutorial/ documentation that one >would go to if one encountered such a gotcha. >("Prominent location" is operationally described as "On the first >screen that opens when you go to this documentation".) Clearer documentation is always better. But given the size of the current documentation this likely won't prevent new users from experiencing the "gotchas" you are describing. Also, keep in mind several of the things that new users seem to trip over are things like thinking in terms of positive reals when Mathematica is designed to treat everything as complex. Issues like this are really related more to mathematics itself rather than Mathematica. >--Or, adding a small number of user-enable-able or disable-able >warning flags that will be displayed if the user issues a command >that may raise one of these gotchas. >Mathematica has, what, about 5000 commands in its vocabulary? Less. Using a tool documented in one of Roman Maeder's books, I get a total of 3442 symbols in the System context of version 7.0 for Mac OS X x86 (64-bit) (February 19, 2009). Some of these of course are constants such as E. Additionally, 580 of these currently have no documentation that is returned by doing ?symbol. So, it seems there is much closer to 3000 things to be covered in the documentation. >Maybe 50 of those commands account for 90% of the gotchas that occur? >Adding maybe 50 such warnings or warning flags wouldn't be a good idea? I am not so sure this would be a good idea. Some of the things being labeled "gotchas" are fairly fundamental. Adding overhead to these things quite likely will impact performance. Also, given the rather fundamental nature of some of these things, there may well be other unintended consequences to other parts of Mathematica. So, changing even a small number of these is likely to entail quite a bit of effort to verify other problems aren't introduced as a consequence of the change.