[ILUG] puzzle of the day - php & apache & phpbb3

[Niall O Broin]

On 28 Jan 2009, at 11:19, Pádraig Brady wrote:

> Niall O Broin wrote:
>> Do you have a setting for    default_charset   in your php.ini? Your
>> HTTP headers say UTF-8 but you have no charet specified in meta tags.
>>
>> Mind you, the corruption you referenced is a bit odd, as it's not
>> happening with non ASCII characters but with random ASCII characters.
>
> It's not random.

It is. The characters being corrupted in the output are random i.e. it  
is not
the more common case one comes across where non ASCII characters are  
not being
correctly handled. You obviously looked more deeply at the problem  
than I did
and came up with this

> For every non ascii character, there is a corresponding �
> at some variable offset from it.
> I.E. whatever is processing the multibyte chars is messing up
> and putting dodgy characters further on in the buffer.
> Note 2 to 3 chars are consumed for each �, so it looks like these
> dodgy characters are being interpreted again as UTF8 and
> converted to the "replacement character" \uFFFD.

which starts to explain what is happening, but doesn't affect the
fact tat the corrupted characters are random (because they are  
characters
AFTER the muti byte character in the output stream, so they could be  
anything.

A tad pedantic, perhaps, but this is ILUG  :-)


Niall