MathGroup Archive 2010

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: Re: (any documentation for) linear syntax?

  • To: mathgroup at
  • Subject: [mg107602] Re: [mg107583] Re: (any documentation for) linear syntax?
  • From: John Fultz <jfultz at>
  • Date: Fri, 19 Feb 2010 03:34:26 -0500 (EST)
  • Reply-to: jfultz at

On Thu, 18 Feb 2010 05:16:32 -0500 (EST), Vince Virgilio wrote:
> On Feb 17, 6:58 am, John Fultz <jfu... at> wrote:
>> This is all a bit unfortunate, because linear syntax can get extremely
>> unwieldy, and it's particularly inside of strings where it becomes most
>> unwieldy due to subtle issues with how the backslash escaping mechanism
>> works within strings. While it's possible to create such forms by hand,
>> we generally don't recommend it, and recommend instead using the various
>> built-in ways of auto- converting when necessary (the documentation
>> points to some of these).
>> If it sounds like I'm not exactly thrilled about how we designed this
>> aspect of the system, that's because I'm not.  There are several gotchas
>> I haven't fleshed out here, and with the hindsight that we now have, we
>> could probably have done this better.  Maybe we will someday.  In the
>> mean time, when you do have to interact with it, it's best to let the
>> system work it out for you.  In the case of Find/Replace, that tends to
>> be pretty simple using the method I described in my first paragraph.
>> Sincerely,
>> John Fultz
>> jfu... at
>> User Interface Group
>> Wolfram Research, Inc.
> What does WRI think of linear format based on Unicode, a la Murray
> Sargent III (and Office 2007)?
> Vince Virgilio

Not speaking for WRI, but just speaking for myself.  I think I'd have to say my 
favorite linear format is still TeX.  I wouldn't want to have to compose or read 
anything non-trivial in either Mathematica's or Word's linear format.  And as a 
software developer, I wouldn't want to have to write either parsing or 
serialization code for either format (yes, I have actually had to do this for 
linear syntax in Mathematica...but it wasn't exactly fun!) .

Computer representations of typesetting map very easily to structured data 
formats, such as Mathematica expressions and XML (or, more specifically, 
MathML).  In general, that's why XML has been so popular among software 
developers.  Highly structured data formats make it much less likely that 
confused and all-too-human software developers will introduce bugs, or QA will 
overlook those bugs, which will turn your Important Data Set into Unreadable
Data Pudding.  It's true that structured data formats can lose some considerable 
human readability.  But TeX, linear syntax, and the Unicode technical note you 
point to also fall down very quickly for non-trivial expressions.

On the other hand, some sort of linear format is going to be with us for a long 
time.  Mathematica still has to support copying and pasting into, for example, 
plain text email.  But I think as it evolves, it will evolve into something more 
structured where there aren't obvious, human-readable/Mathematica-interpretable 
linear representations.  In fact, such an evolution has already been happening 
(while maintaining backward compatibility, of course).


John Fultz
jfultz at
User Interface Group
Wolfram Research, Inc.

  • Prev by Date: Re: Re: Re: May we trust IntegerQ
  • Next by Date: Re: Re: (any documentation for) linear syntax?
  • Previous by thread: Re: (any documentation for) linear syntax?
  • Next by thread: Re: Re: (any documentation for) linear syntax?