1197113 - <input type=number> with step 0.01 treats "." as a grouping separator instead of the decimal separator

Status: RESOLVED INCOMPLETE

(Core :: DOM: Core & HTML, defect)

Description

[enclave.pevik]

User Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:42.0) Gecko/20100101 Firefox/42.0
Build ID: 20150819004006

Steps to reproduce:

"Mozilla/5.0 (Windows NT 10.0; WOW64; rv:42.0) Gecko/20100101 Firefox/42.0"
Developer Edition

I've created input as follows inside of the form:
<input type="number" min="0.01" max="9999999999999.99" step="0.01" />


Actual results:

I console.log the value in javascript on form submission. Here are the results:
When I type in "1.23" or "1,23" -> OK, results in 1.23 string value, BUT:
When I type in "1.234" -> I got value of "1234" in console! [ERROR]
When I type in "1,234" -> validation works and FireFox won't allow me to send the form.
When I type in "1.2345" or "1,2345"-> validation works and FireFox won't allow me to send the form.


Expected results:

1) FireFox shouldn't convert the value from 1.234 to 1234
2) FireFox shouldn't allow me to type in 1.234 because of the "step" attribute

Comment 1

[Jonathan Watt [:jwatt]]

What is your locale? (If you go to the URL about:config then what is the value of the general.useragent.locale pref?)

Comment 2

[enclave.pevik]

It's set to en-US

Comment 3

[Jonathan Watt [:jwatt]]

Attachment: testcase

Comment 4

[Jonathan Watt [:jwatt]]

Attachment: testcase

Comment 5

[Jonathan Watt [:jwatt]]

And what does the attached testcase alert for you?

Comment 6

[arni2033]

I have russian locale, and I confirm results in original report. Also, I agree with expectation (1)

(In reply to Jonathan Watt [:jwatt] from comment #5)
> And what does the attached testcase alert for you?
About your testcase: input 1,234 - results in "1234" alert.   1.234,5 results in "1234.5".
I suppose that's not how it should work with [lang="en"] (something's wrong with lexical interpreter)
Therefore I'm confirming this bug w/o hesitation
I also don't know why devs ever allowed to type group separator since it's confusing between locales

Comment 7

[arni2033]

I'm very sorry for my huge mistake. Everything (discussed in this report) WFM as expected on russian and english localizations of Aurora. I mixed it up with bug 1208923 I just reported.

I agree that the reporter obviously encountered this on a page with   [lang="{comma-language}"]   since "1,23" and "1.23" both passed the validation.

Comment 8

[enclave.pevik]

(In reply to Jonathan Watt [:jwatt] from comment #5)
> And what does the attached testcase alert for you?

Now it works when I type 1.234 (alerts 1.234), but when i type in 1,234 it alerts 1234. So it looks like language-specific problem.

I don't think this should be language-specific, since "." and "," are both valid decimal separators in many countries. I've never read the input specs by Consortium, but I thought that defining input as number both "." and "," should be valid decimal separator chars.

Comment 9

[Jonathan Watt [:jwatt]]

(In reply to enclave.pevik from comment #8)
> (In reply to Jonathan Watt [:jwatt] from comment #5)
> > And what does the attached testcase alert for you?
> 
> Now it works when I type 1.234 (alerts 1.234), but when i type in 1,234 it
> alerts 1234. So it looks like language-specific problem.

The lang="en" on the <input> is expected to behave in exactly this way. In English, "." is the decimal separator, while "," is the thousands separator.

> I don't think this should be language-specific, since "." and "," are both
> valid decimal separators in many countries.

Which countries? In the locales that I'm aware of one is the decimal and one is the thousands separator. I'm not aware of a locale in which *both* are interchangeably usable as the *decimal* separator. I'm no expert on this though, hence the question.

Comment 10

[Jonathan Watt [:jwatt]]

Marking "INCOMPLETE" pending a reply to my question in comment 9. Otherwise this seems to be invalid.