12385 - CSS idents can't start with "-"

Status: RESOLVED INVALID

(Core :: CSS Parsing and Computation, defect, P3)

Description

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

DESCRIPTION:  CSS2 identifiers can't start with dash (0-9 and - are the
characters that can't start an identifier but can be in the middle).  I know
you're using -moz to begin identifiers that represent extensions, but (unless
this will be changed in CSS3), you shouldn't allow classes, etc., that begin
with a -.  That is:

.-d { font-weight: bold; }

should be ignored.

STEPS TO REPRODUCE:  Load above URL.

ACTUAL RESULTS: Fifth bulleted item has two words in bold.

EXPECTED RESULTS: Fifth bulleted item should not have any words in bold.

DOES NOT WORK CORRECTLY ON:
 * Linux, apprunner, 1999-08-20-13-M10

Comment 1

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

ALSO DOES NOT WORK CORRECTLY ON:
 * Windows, apprunner, 1999-08-24-09-M10

Comment 2

[Peter Linss]

CSS3 allows idents to start with dashes (explicitly to allow the extension
mechanism). Need to evaluate if this should apply to class names (and a few
other places where ident is allowed.

Comment 3

[Christine Hoffman]

Verifying this LATER.

Comment 4

[Pierre Saslawsky]

Reassigning peterl's latered bugs to myself.

Comment 5

[Christine Hoffman]

Is this still latered? If not, please remove resolution. If not, I will verify
it again.

Comment 6

[Pierre Saslawsky]

Yep, still latered.

In case we (er... when we) look into it again, Ian posted some useful info about
errata to the CSS syntax under bug 18936.

Comment 7

[Christine Hoffman]

Verified LATER.

Comment 8

[Hixie (not reading bugmail)]

Migrating from {css1} to css1 keyword. The {css1}, {css2}, {css3} and {css-moz}
radars should now be considered deprecated in favour of keywords.
I am *really* sorry about the spam...

Comment 9

[Hixie (not reading bugmail)]

Reopening and moving to Future...

Comment 10

[Tobias Burnus]

I think it would be nice, if it would have an milestone (instead of Future)
assigned since Mozilla claims to support CSS fully.
See also:
http://www.w3.org/Style/CSS/Test/current/sec15.htm
and specs: http://www.w3.org/TR/REC-CSS2/syndata.html#q4

Comment 11

[hidden]

Attachment: testcase

Comment 12

[hidden]

added testcase

Comment 13

[Kurt Weinschenker]

Tried http://www.w3.org/Style/CSS/Test/current/sec15.htm on Mac OS 9.0.4,
Mozilla build 2000100208-M18, still doesn't work.

(CSS1 Test Suite:  1.5 ID as selector)

Comment 14

[Hixie (not reading bugmail)]

Netscape's standard compliance QA team reorganised itself once again, so taking 
remaining non-tables style bugs. Sorry about the spam. I tried to get this done 
directly at the database level, but apparently that is "not easy because of the 
shadow db", "plus it screws up the audit trail", so no can do...

Comment 15

[Daniel Glazman (:glazou) (not active in Mozilla any more)]

Changing URL to my test suite since my publisher modified its web site
architecture w/o notification :-(

Comment 16

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Assigning pierre's remaining Style System-related bugs to myself.

Comment 17

[Boris Zbarsky [:bzbarsky]]

Has the WG decided anything here yet?

Comment 18

[Karl Ove Hufthammer]

> Has the WG decided anything here yet?

I don't know, but they have at *not* allowed the
'-' character in the CSS 2.1 WD, released ~a week ago
<URL: http://www.w3.org/TR/2002/WD-CSS21-20020802/grammar.html >.

Comment 19

[Boris Zbarsky [:bzbarsky]]

Yes... but this is not a CSS2 issue.  This is a question of what CSS3 will say
(since CSS2 also does not have the -vendor-ident mechanism in place).

Comment 20

[Anne (:annevk)]

CSS2.1 does: <http://www.w3.org/TR/2004/CR-CSS21-20040225/syndata.html#q4>

Comment 21

[Boris Zbarsky [:bzbarsky]]

Yeah, at this point this looks invalid.

Comment 22

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

"Only properties, values, units, pseudo-classes, pseudo-elements, and at-rules
may start with a hyphen (-); other identifiers (e.g. element names, classes, or
IDs) may not." (4.1.3)

Comment 23

[Anne (:annevk)]

To summarize: this is a valid bug, since Mozilla supports something it should
not support (.-d).

Comment 24

[Justin Wood (:Callek)]

hmm iirc recently we had a problem with calendar class names (which started with
'-' 

iirc we warn about it in debug builds, perhaps we can make the warning into an
error?

Comment 25

[Gérard Talbot]

Latest CSS 2.1 WD (06 November 2006) suggests that IDs can start with an hyphen (-) and then may be followed by an alphabetic character [A-Za-z] or may then even be followed by an underscore (_).

{
In CSS, identifiers  (including element names, classes, and IDs in selectors) can contain only the characters [a-z0-9] and ISO 10646 characters U+00A1 and higher, plus the hyphen (-) and the underscore (_); they cannot start with a digit, or a hyphen followed by a digit.
}
CSS 2.1, Working Draft 06 November 2006, Section 4.1.3 Characters and case
http://www.w3.org/TR/CSS21/syndata.html#q6

{
ident		-?{nmstart}{nmchar}*
nmstart		[_a-z]|{nonascii}|{escape}
}
CSS 2.1, Working Draft 06 November 2006, G.2 Lexical scanner
http://www.w3.org/TR/CSS21/grammar.html#q2

To make things worse,
CSS 2.1, Working Draft 06 November 2006, section G.3 Comparison of tokenization in CSS 2.1 and CSS1 at
http://www.w3.org/TR/CSS21/grammar.html#tokenizer-diffs
says 
{
In CSS1, a class name could start with a digit (".55ft")
}
while CSS 1 spec says
{
in CSS1, selectors (element names, classes and IDs) can contain only the characters A-Z, 0-9, and Unicode characters 161-255, plus dash (-); they cannot start with a dash or a digit;
}
http://www.w3.org/TR/REC-CSS1#css1-conformance

Finally, HTML 4.01 claims that
{
ID and NAME tokens must begin with a letter ([A-Za-z]) and may be followed by
any number of letters, digits ([0-9]), hyphens ("-"), underscores ("_"), colons
(":"), and periods (".").
}
http://www.w3.org/TR/html4/types.html#type-id

Did I miss something? 
I'm voting for this bug :)

Comment 26

[Gérard Talbot]

Example:
<p id="-asdf">

Both the HTML 4.01 spec and W3C HTML validator say this attribute value is invalid

but

#-asdf { color: red }
is fine and ok with latest CSS 2.1 WD and with the W3C CSS validator

Comment 27

[Gérard Talbot]

Complete testcase for this bug:

http://www.gtalbot.org/BrowserBugsSection/MSIE7Bugs/CSS1ForwardCompatibleParsing.html

Comment 28

[Boris Zbarsky [:bzbarsky]]

We pass that testcase, no?

Comment 29

[Gérard Talbot]

If we meticulously follow and comply with the latest CSS 2.1 (6 nov. 2006) WD, then yes, we pass that testcase.


I wish all CSS 2.1 WD would not have introduced exceptions and unneeded complexity and just carried on with the CSS1 and simple 
"selectors (element names, classes and IDs) (...) can not start with dash or digit".
Those kind of new CSS 2.1 exceptions encourage and promote invalid markup code as invalid attribute specifications are to be honored nevertheless in CSS 2.1.

Comment 30

[Ray Kiddy]

FYI: There is a CSS 2.1 test that fails that is, I believe, from this bug.

The source has:

  .one { color: green; background: white; }
  .-ident, .one { color: red; }

  .two { color: green; background: white; }
  #-ident, .two { color: red; }

In both cases, the text should be green, but it is red.

http://www.w3.org/Style/CSS/Test/CSS2.1/current/t040103-ident-03-c.htm

Looking at:
http://www.w3.org/Style/CSS/Test/CSS2.1/current/t040103-ident-00-c.htm
...
http://www.w3.org/Style/CSS/Test/CSS2.1/current/t040103-ident-13-c.htm

All the others pass.

Comment 31

[Gérard Talbot]

Ray,

The previous and current/latest [July 19th 2007, CR] CSS 2.1 spec says:
"in CSS, identifiers  (including element names, classes, and IDs in selectors) can contain only the characters [a-z0-9] and ISO 10646 characters U+00A1 and higher, plus the hyphen (-) and the underscore (_); they cannot start with a digit, or a hyphen followed by a digit."
http://www.w3.org/TR/CSS21/syndata.html#value-def-identifier

Also, current (and previous) CSS 2.1, Section 4.1.1 Tokenization
http://www.w3.org/TR/CSS21/syndata.html#tokenization
still gives
ident 	[-]?{nmstart}{nmchar}*
nmstart [_a-z]|{nonascii}|{escape}

So, they can start with an hyphen as long as their next character is not a digit. E.g.: 

1-
id="-_" 
#-_ {color: red;}
is perfectly valid CSS 2.1

2-
id="-foo"
#-foo {color: red;}
is perfectly valid CSS 2.1

3-
class="-bar"
.-bar {color: red;}
is also perfectly valid CSS 2.1

and so is -ident since "i" is not a digit.

This is the CSS validator report according to CSS 2.1 profile of the t040103-ident-03-c.htm webpage:
http://jigsaw.w3.org/css-validator/validator?uri=http%3A%2F%2Fwww.w3.org%2FStyle%2FCSS%2FTest%2FCSS2.1%2Fcurrent%2Ft040103-ident-03-c.htm&warning=2&profile=css21&usermedium=all

As far as I can see, both sentences in 
http://www.w3.org/Style/CSS/Test/CSS2.1/current/t040103-ident-03-c.htm
should be red, not green. That test needs to be updated.

Comment 32

[Gérard Talbot]

As far as I'm concerned, 
- this bug should be resolved as WORKSFORME, 
- the summary should be updated so that we don't mislead bug reporters 
- I. Hickson should also be informed that one of his test 
CSS2.1/current/t040103-ident-03-c.htm
may not be correct.

Comment 33

[fantasai]

The restriction on starting hyphens in identifiers was removed.
The resolution was for issue 167 in issues-3, in response to
  http://lists.w3.org/Archives/Public/www-style/2006Mar/0068.html
The discussion is minuted in
  http://lists.w3.org/Archives/Member/w3c-css-wg/2006JulSep/0006.html
where it basically says there's interop on supporting hyphens at the beginning of class names and the restriction is confusing anyway.

Marking INVALID.