Please explain why do you think it should succeed if the data is invalid.

If the data is NOT valid, FALSE should be returned, I guess.
But actually it returns TRUE.
Am I wrong or missing your point?

It is expected behavior because 
0x00,0xa1 is null byte + valid ShiftJIS character.
mb_check_encoding should be used to detect invalid 
or corrupted multibyte characters.

Just read bug report again.

No one says 0x00,0xa1 is invalid character in ShiftJIS.

Someone disagrees, Rui.. :)

> No one says 0x00,0xa1 is invalid character in ShiftJIS.
I didn't say that.

0x00+0xa1 is valid byte sequence in Shift_JIS sequence.
A character in Shift_JIS encoding is encoded in either single byte 
or double byte.
In this case, the byte stream is reconigzed as two character,
a null byte and a comma character in Katakana(0xa1) 
 
see: http://hp.vector.co.jp/authors/VA013241/misc/shiftjis.html

> 0x00+0xa1 is valid byte sequence in Shift_JIS sequence.

I know it.
But 0x00+0x81 is invalid sequence in Shift_JIS.
Then, why below statement returns "bool(true)" ?

var_dump(mb_check_encoding("\x00\x81", "Shift_JIS"));

Read bug report again, please.

Did you read it Rui? (why do your reports end up as 'Analyzed' all the time? :)

Sorry for delaying response.

0x00,0x81 is also valid byte sequence in Shift_JIS
because 0x81 is a valid first byte of a double-byte 
JIS X 0208 character.

See: http://en.wikipedia.org/wiki/Shift_jis

We cannot decide the byte stream is valid or 
invalid because the last byte of byte stream (0x81)
is a valid first byte of double-byte character.
In this case, true (valid) will be returned.

The byte stream including a valid first byte +
a invalid second byte returns false.

For example,

var_dump(mb_check_encoding("\x81\x00", "Shift_JIS"));

returns false (invalid).

It is because 0x81 is valid first byte of a double-byte
JIS X0208 character, but, 0x00 is invalid second byte of
a double-byte JIS X0208 character.

And, 
0x00, 0xe3 in UTF-8, it is also 
valid byte sequence (a null byte + first byte of 
a three-byte UTF-8 character).

See: http://en.wikipedia.org/wiki/UTF-8

0x00, 0xe3 is a valid byte sequence in UTF-8 but by itself is not a valid UTF-8 string (it's missing two bytes).

The function is documented as checking the validity of a string so it should return false for this case. If the function is only supposed to validate byte-streams then the documentation should be fixed.

This bug has been fixed in CVS.

Snapshots of the sources are packaged every three hours; this change
will be in the next snapshot. You can grab the snapshot at
http://snaps.php.net/.
 
Thank you for the report, and for helping us make PHP better.

It is a documentation problem, and it is already fixed in CVS.

I guess the problem is not only in the document.

var_dump(mb_check_encoding("\x00\xE3","UTF-8"));
=> bool(true)        may be checking validity of "byte streams"

var_dump(mb_check_encoding("\xE3", "UTF-8"));
=> bool(false)       may be checking validity of "string"

# I hope that this function checks validity of "string", not "byte streams" (but this is just my opinion).