I try to avoid using spaces in filenames and URLs. They're great for human readability, but they're remarkably inconvenient in computer resource locators:
This_is_a_single_word
but this-is-multiple-words
ā Jeff Atwood
Although the second statement is true for almost all Windows applications; the first statement is NOT always true, especially for Microsoft Office products (Word, Outlook) even Windows text box dialogs.
You can test this by using the above examples in a Word document, Outlook email, text box, etc.; searching for one individual token of the entire expression, e.g. āwordā. In these applications, the search will find āwordā in both of the hyphenated AND underscored versions of the expression.
Also, you can Ctrl-Left-Arrow/Ctrl-Right-Arrow to the first last character of each token in both versions. Whereas applications that treat the underscored version as a single āwordā, Ctrl-Left-Arrow/Ctrl-Right-Arrow will advance only to the first last character of the entire expression.
By my reading, this interpretation (dash as a word break character but not underscore) is continued in the Unicode world. It defines the rules for word boundaries here:
http://www.unicode.org/reports/tr29/tr29-9.html#Word_Boundaries
(Just for reference, in unicode terminology, the underscore is called a āLOW LINEā, and the dash is called a hyphen. There are several hyphen characters, see the link above.)
ithe first statement is NOT always true, especially for Microsoft Office products (Word, Outlook) even Windows text box dialogs./i
As a developer, this always drives me nuts. I expect a text box to treat underscores and dashes like my IDEs treat them: an underscore as a letter, and a dash as āpunctuationā. In programming of course, this makes sense. An underscore is not an āoperatorā in the context of many (most?) common programming languages, unlike the dash (minus), or whitespace (symbol separator).
However, an underscore really has no official place in the English language. It was created for the typewriter, as a way of underlining. Itās word-separation usage only came with the advent of 2GLs. So confusion is understandable, if not really acceptable, when dealing with word-processing functionality.
At any rate, it would be nice if everyone treated it the same way. At least within a given context.
Underscores are also ugly and hard to type quickly, whereas dashes are easy and the same width as a space, or only slightly longer. Underscores at 2-3 spaces wide make words look disconnected, even if theyāre otherwise less obtrusive.
Of course you could simply use + and ban the use of it in filenames, with appropriate server support. (Or be prepared to use %2B.)
Iām-more-concerned-about-which-one-is-more-readable.
Iām_more_concerned_about_which_one_is_more_readable.
I tend to prefer the underscore because itās below the letters, as if in an underline. But I donāt think thereās a right answer. Just preference.
Another difference between underscores and dashes: very often, you canāt see underscores when the words are underlined (as in most hyperlinks). Sometimes this is an advantage, but in most cases it isnāt.
2 little notes from a unixish perspective (I thought Iād share because Iām reading your blog to get the windowsish perspective):
\w is not portable regex, but rather from the preg (perl regex) set.
Also, spaces in file names, esp. as arguments to shell scripts, are easily lost because once inside the script, evalling a string more than once makes it āfall apartā, quoted or not.
The solution there is the magic variable ā$@ā (quotes included) which evals to all positional arguments, properly quoted one by one.
I like this one better:
IāMMOR~1
I remember a quote somewhere, something like :
8 chars ought to be enough for anybody. 
nah, sorry to low ā¦
We switched from _ to - a few years ago for usability reasons: some folks donāt see an _ in a URL and assume itās a space.
Hereās another vote for ā.ā as a word separator in filenames (does UNICODE call that a FULL STOP? - actually, it does).
I have no in-depth analysis for why, itās just what I have always used to avoid the evils of spaces in filenames.
this-is-a-long-filename
this.is.a.long.filename
this_is_a_long_filename
I still like it better.
This was really useful. Iām working a file to convert search friendly urls into a keyword search on my website. I was going to use underscores but now Iām going to use dashes or underscores. Thanks.
I, too, hate the spaces! You donāt know how badly Iāve wanted to change āProgram Filesā to ProgramFiles or Program_Files.
That being said, what about camel/pascal casing? Seems like that would be another option for this discussion. I wonder how the readability changes?
The space/dash/underscore self-deliberation comes up frequently as Iām ripping CDās to mp3s and trying to get a good filename. Iāve resolved to using _ for spaces and ā for a delimiter.
I guess I would argue that the _ makes more sense as a space, but maybe thatās because thatās what Iāve always used.
I agree that itās nice to adhere to a standard, but if there are good reasons to throw the standard out the window, why not consider it? Not that Iāve come up with good reasonsā¦
renaming āProgram Filesā can break some archaic installers / updates so I wouldnāt risk it
I personally simply add a Junction (Symbolic Link) that maps Program Files to Program_Files (so they point at the same place on a NTFS disc) and I can then refer to it in either way. Very handy
Interestingly iāve heard Vista will be dropping all the two word directoring and moving from Program Files - Programs, My Documents - Documents etc.
p.s. util to help create junctions: http://www.sysinternals.com/Utilities/Junction.html
Jeff, thatās terrible advice. If programmers donāt use the same characters in their filenames that their users use then how is software ever going to work properly?
W3Cās CSS validator is broken because somebody didnāt think to check it with a URL that has a %20 in it: http://www.kirit.com/W3C%27s%20CSS%20validation%20service
Microsoftās Response.Redirect() is broken because it canāt work out which bit of a URL is which, also due to sloppy encoding practiced by most web developers: http://www.kirit.com/Response.Redirect%20and%20encoded%20URIs
And again a problem with encoding means that 404 handlers on IIS are broken too: http://www.kirit.com/Errors%20in%20IIS%27s%20custom%20404%20error%20handling
And because youāve reverse engineered Googleās software doesnāt seem a good reason either. Sooner or later theyāll change their implementation and then where will you be? If Google considers a hyphen as a space then Google is broken. Sooner or later theyāll fix that (one hopes).
One comment suggests ā+ā, like Technorati uses in tags? Thatās even more broken ā the ā+ā is used as a space substitute in query strings but NOT in file specifications, a bug in nearly every URL parser Iāve found.
āIf you use an underscore ā_ā character, then Google will combine the two words on either side into one word.ā
Personally, I feel this is Googleās problem, not mine. But thatās just my opinion. I will only go so far to accomodate Google. (And it shows; their cache of my site is a mess, because my CMS throws insane session cookies at Googlebot, in the URLās querystring. I havenāt gotten around to fixing it yetā¦ see previous comment.)
Hereās another vote for ā.ā as a word separator in filenames (does UNICODE call that a FULL STOP? - actually, it does).
One problem with period is that it makes it difficult to figure out where the file extension begins, eg:
this.is.a.html.file.html
Spaces in URLs are bad because they have to be replaced with that ugly unreadable %20 notation. But what exactly is wrong with spaces in file names? As long as you donāt need to share a file with backwards Unix systems that donāt understand file names with spaces I donāt see the problem. Where do you ever enter a file name?
In a file selection dialog. Standard Windows file dialogs donāt care about spaces. No quotes are necessary.
On the command line. Auto-completion automatically puts quotes around your file name as necessary.
In a programās source code. Strings must be surrounded by quotes anyway, ergo spaces are not a problem.
In some text storage facility, such as the registry or an XML file or whatever. Any well-designed storage format respects embedded spaces, so once again no problem.
The only problematic situation I can come up with (other than sharing with Unix systems) are batch files. Thatās the only time you have to consciously remember to use quotes for file names with spaces. And even thatās no longer true when you upgrade to Windows PowerShell!
Cāmon guys! Either computers work for you or you work for them.
Stick with spaces.
There_is_a_reason_that_we_donāt_write_our_sentences_like_this. There-is-also-a-reason-that-we-donāt-write-our-sentences-like-this. And.this.is.a.joke!
Iāll let you hardcore filename worshipers use dashes, dots, and underscores. Iāll use spaces.
You forgot the Wiki style: no spaces!
ThereIsAReasonThatWeDonātWriteOurSentencesLikeThis
Which I suppose is what other posters were referring to with āCamel Case styleā, but I think of this as Wiki-styleā¦