MathGroup Archive 2008

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

Search the Archive

Comments on the .m file editor

  • To: mathgroup at
  • Subject: [mg87833] Comments on the .m file editor
  • From: Will Robertson <wspr81 at>
  • Date: Fri, 18 Apr 2008 02:40:23 -0400 (EDT)


As essentially a new user of Mathematica since v6, I'd like to share a
couple of comments about using Mathematica itself to write .m packages
directly. I'm only using v6.0.1 because I can't afford the bug-fix
update right now, so a couple of these comments may no longer be

Firstly, the positives: It's great to be able to write code in an
editor that explicitly understands the syntax. Brace matching and
selection are good, and syntax colouring is excellent.

Being able to write comments in outline mode is also great; especially
with "input cells" that aren't part of the package but allow you to
execute code; this makes testing and code exposition very convenient
because it can be done inline with the code itself.

Now, the negatives. The first complaint is more ignorance than
anything: I've got no idea how to use the debugger. It's great that
it's there, but I don't understand how it's supposed to be used. Are
there any tutorials on this feature?

Next: the editor *really* needs to make up its mind whether it's going
to support "nice" Mathematica symbols or not. It's ridiculous that
typing "-> " gives you a nice arrow, but then closing and re-opening
the file gives you the literal ascii. Worse, typing a complex
expression involving Greek letters and subscripts and accents looks
fine to start with but then turns into an unreadable mess when
re-opening the package.

Since Mathematica can understand the FullForm in package files, I can't
see the harm in displaying it nicely in the .m file editor. Similarly,
the lack of nice spacing in the typesetting of the code is a shame and
makes packages look uglier -- usually requiring extra whitespace (and
effort) to be readable.

Finally, the editor doesn't like code that has comments inserted
mid-way through. For example, paste this into a file test.m:

(* ::Package:: *)
(* ::Text:: *)
(*some explanation*)

Open it in Mathematica, then select all, cut, and paste (this procedure
emulates you typing in those lines manually). Close and reopen the file
and you'll see that Mathematica has inserted two blank lines in the
code. Yuck.

Furthermore, this sort of construction kills syntax colouring, which is
a big shame.
It also breaks the "Run Package" button, which tries to evaluate the
package cell by cell (instead, it should simply execute `<<"test.m"`).

I don't think that cells should be restricted to contain self-contained
chunks of code when large packages deserve better comments than an
essay right before the Module with inline comments like (* we're doing
stuff here *). If those sorts of comments was the only way to document
the code then we may as well not even bother with cells, to be frank.


To sum up: the .m file editor is pretty neat but two main additions
would make it much better: explicit support for breaking cells up with
comments whereever you like (including syntax checking and colouring
across cell breaks); and typesetting as in the the notebook editor
(including the extended symbols). In the very least, it shouldn't
interpret typeset symbols unless it intends to preserve them. The
typesetting doesn't affect anyone trying to edit the thing in plain
text and makes the whole thing much nicer to use.

After all, without nice typesetting we may as well go back to notebooks
with initialisation cells, which have even less ability to be commented
nicely by splitting up cells partway through.

Any comments from others also using the editor?

Best regards,
Will Robertson

  • Prev by Date: Re: Re: EdgeRenderingFunction to produce edge
  • Next by Date: Re: A kernel, multiple notebooks, and Global?
  • Previous by thread: Built-in DatePlus and DateDifference are much slower than DaysPlus
  • Next by thread: Re: Comments on the .m file editor