Very Computer

Does anyone know of any usability studies into wildcards, and wildcard
standards? Had a look on the web and couldn't find anything apart from
'novice users don't understand or use wildcards' and 'expert users only use
basic wildcards'.

Best,

- h

I have no information, but a strong recommendation: if you DO use
wildcards, use the standard notation. The same applies to Boolean
logic. Don't make up your own "easier" system, as eBay and many search
engines have done. That gives users the worst of both worlds: both
novice AND experienced users must learn how to use them, and
experienced users must remember when to use your non-standard system
(along with whatever other ones they're stuck with).

My mail address is jsachs177 at earthlink dot net.

Quote:> I have no information, but a strong recommendation: if you DO use
> wildcards, use the standard notation.

And just what *is* the standard notation? As far as I can see, there are
'standards' where '%' is used for 0 or more character matches and '_' is
used for a single-char match, and others where '*' is used for 0 or more
characters and '?' is used for a single char match. And then there are the
other examples where other forms are used.

Best,

- h

I'd have to second that. And for all that I'm personally a fan of
regexp, I think that outside of specialized applications where detailed
searching is a constant and central activity, you'd be best using glob-
style wildcards for your matching.

Beyond that, I would argue that wildcards should be optional. Pick a
reasonable default behavior, and allow users to use wildcards to refine
that behavior if they choose. What's 'reasonable' here depends a lot on
the data that's being searched and the pattern of use.

--S

Whatever Google is using, use that.

    --bks

The standard notation is that used by the Microsoft file system,
unless you're designing for a specific environment where a plurality
of users who use computers at all are most comfortable with something
else.

My mail address is jsachs177 at earthlink dot net.

<nitpick>
   It's not the filesystem that implements wild cards, it's
   the command processor (in DOS, typically Command.COM)
</nitpick>

Also:

- The DOS wildcard mechanism has evolved over time. The original was
  one of the worst ever conceived (hm, a quick search learns that
  later ones may even be worse; for instance, see
     <http://codeguru.earthweb.com/files/CAdvFind.html>
  Please allow multiple occurrences of '?' and/or '*' in a search
  term
- Should we really call this DOS wildcards:
  - I think the use of '*' and '?' in regular expressions predates
    DOS (possibly by a decade or more)
  - Does the average user know about DOS nowadays?

Reinder

[Jonathan Sachs]

| The standard notation is that used by the Microsoft file system,
| unless you're designing for a specific environment where a plurality
| of users who use computers at all are most comfortable with something
| else.

hogwash.

Microsoft uses (as is its wont) its own syntax loosely (very) based on
regular expression syntax.  regular expressions come from computational
theory.  there are two widely used standards, POSIX 1003.2 is one and
the programming language Perl provides another one (which seems to
change with each new release).

if only Microsoft could use the same basis for its command interpreter
and its editing environments.

--
Rolf Lindgren                                            http://www.roffe.com/

Quote:>| The standard notation is that used by the Microsoft file system...

>hogwash... there are two widely used standards, POSIX 1003.2 is one and
>the programming language Perl provides another one...

Do you really expect POSIX or Perl to be taken seriously as a de facto
standard with a better claim than Microsoft Windows?

Quote:>...(which seems to change with each new release)

Which means that it is not a standard at all.

My mail address is jsachs177 at earthlink dot net.

[Jonathan Sachs]

| Do you really expect POSIX or Perl to be taken seriously as a de facto
| standard with a better claim than Microsoft Windows?

yes, as a matter of fact I do - for any application more advanced than
the Windows shell.  

[...] (on Perl's regexp)

| Which means that it is not a standard at all.

certainly.  as a matter of fact, for any project that would seem to
require more knowledge of Perl than can be obtained in half an hour,
there is another language that is more appropriate.

--
Rolf Lindgren                                            http://www.roffe.com/

Quote:>> Do you really expect POSIX or Perl to be taken seriously as a de facto
>> standard with a better claim than Microsoft Windows?

>yes, as a matter of fact I do

Then you will be disappointed.

I think you have misinterpreted the term "standard" as it applies to
this thread. It does not mean "a design which authorities have agreed
to implement in products, which will therefore be learned and used by
users." It means "a design which a significant group of users already
understand, which will therefore be used if implemented."

I will guarantee you that apart from computer professionals, a vast
majority of users haven't a clue whether POSIX and Perl are software
components, hardware components, or rock groups. Nor can they be
induced to care.

My mail address is jsachs177 at earthlink dot net.

<nitpick^2>
 the command processor in dos didn't do any wildcard expansion.

 the wildcard strings were passed to the .exe's for them to interpret.

 hence no standard, each exe/com did what it wanted.

 full story in dr dobbs journal, about 18 months ago, al steven's column.  
</nitpick^2>

martin

Since my first, insufficiently clear post, I have tried to emphasize
that I am talking about a de facto standard. How the standard was
implemented really is not relevant.

As a point of information, though, DOS did have standard wildcard
processing in the form of a set of system calls. Thus applications
which supported wildcard processing at all had to go to considerable
trouble to do so in a non-standard way.

My mail address is jsachs177 at earthlink dot net.