1419958 - Firefox 57 shows question marks instead of letters with diacritics with latin-1 encoded page

Status: VERIFIED INVALID

(Firefox :: Untriaged, defect)

Description

[Rogério Brito]

Attachment: Screenshot at 2017-11-22 20-14-58.png

User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:57.0) Gecko/20100101 Firefox/57.0
Build ID: 20171115002005

Steps to reproduce:

Visited https://www.ime.usp.br/~rbrito/


Actual results:

The text is full of question marks. I'm attaching a screenshot of what I see.


Expected results:

It should display all the accented characters.

Comment 1

[Kohei Yoshino [:kohei]]

I can confirm the issue, but it's an issue of the site itself: the Content-Type HTTP header says UTF-8 while the page is written in iso-8859-1. It's recommended to always use UTF-8 for page authoring.

https://www.w3.org/International/questions/qa-html-encoding-declarations
https://github.com/whatwg/html/pull/3091

Comment 2

[Rogério Brito]

Sure, I agree that everything should be UTF8.

But it was working before (was this a bug with earlier Firefox versions?) and I didn't touch the page in question for about the last decade...

I will change the page to be UTF-8, but I expected this to be rendered as the author of the page requests... Since this bug is closed as invalid, I am powerless, though...

Comment 3

[Rogério Brito]

Thanks for the feedback and not ignoring the issue, though.

Comment 4

[Matthias Versen [:Matti]]

The server sends this header which can be seen with http://websniffer.cc or the integrated Firefox developer tools: 
Content-Type:	text/html; charset=UTF-8 
The http charset header always overrides any charset declaration in the document. 
A good reason for that is that you already have to know the used charset if you want to parse the headers inside the document.

The behavior didn't change in a long time as far as I know.
I bet that that your server didn't send the charset declaration in the past.

Comment 5

[Rogério Brito]

(In reply to Matthias Versen [:Matti] from comment #4)
> The server sends this header which can be seen with http://websniffer.cc or
> the integrated Firefox developer tools: 
> Content-Type:	text/html; charset=UTF-8 
> The http charset header always overrides any charset declaration in the
> document. 
> A good reason for that is that you already have to know the used charset if
> you want to parse the headers inside the document.
> 
> The behavior didn't change in a long time as far as I know.
> I bet that that your server didn't send the charset declaration in the past.

I honestly don't know. When I wrote that page more than 10 years ago, I tried to use what I learned were the correct practices and I checked the pages against the (at that time, very useful) W3C validator.

The web server is not under my control (it is my Institute's web server). I just changed all the pages to have UTF-8 encoding, so you may not see the bug happening again.

I guess that this is too oddball to care and I guess that this bug report can be closed.


Thanks once again,

Rogério.