Chrome Version       : Version 50.0.2661.102 (64-bit)
URLs (if applicable) : not applicable
Other browsers tested:

     Safari: not tested
    Firefox: not tested
         IE: not tested

What steps will reproduce the problem?
(1) open local file

What is the expected result?

Open a file and show it linted in XML.


What happens instead?

============
This page contains the following errors:

error on line 18579 at column 25: Input is not proper UTF-8, indicate encoding !
Bytes: 0xE2 0x80 0xA2 0x20
Below is a rendering of the page up to the first error.
============

$ md5sum file.xml 
6b0938f815c1e86e5b5f4e29b6fe9040  file.xml

$ xmllint --noout file.xml 


I inspected the file manually and indeed it looks like valid unicode
============

$ head -n1 file.xml
<?xml version="1.0" encoding="UTF-8" ?>


$ grep -b 'ontwikkelen van maatwerk-applicaties' file.xml 
2353958:  <description><![CDATA[<p>• ontwikkelen van maatwerk-applicaties in php e
...



$ hexdump -s 2353958 -n 120 -v -e '"|" 32/1 "%_u" "|\n"' file.xml 
|  <description><![CDATA[<p>e280a2 o|
|ntwikkelen van maatwerk-applicat|
|ies in php en nodejs<br />e280a2 im|
|plementeren van enterpri|
============
0xe2 0x80 0xa2 looks like valid unicode for UTF8 ('BULLET' (U+2022)). However chrome in its error says:

Bytes: 0xE2 0x80 0xA2 0x20 (note the 'SPACE' (U+0020) after the BULLET)


In the bug reporting system, there is a max attachement file of 10M. I tried to create a minimal working example but I failed (I attach two tests that work OK - they do NOT show the problem). The bug is reproduced only by a 90M xml file (I understand why my claim is frustrating :( ). I can send the file if needed in another way.

 

Deleted: test2.xml 1.8 KB

test2.xml 1.8 KB View Download

Deleted: test.xml 144 bytes

test.xml 144 bytes View Download