We've fixed it be commenting line 421 of ext/standard/html.c:

420-    { '\'', "'",       6,      ENT_HTML_QUOTE_SINGLE },
421:/*  { '\'', "'",        5,      ENT_HTML_QUOTE_SINGLE }, */

Thank you for taking the time to write to us, but this is not
a bug. Please double-check the documentation available at
http://www.php.net/manual/ and the instructions on how to report
a bug at http://bugs.php.net/how-to-report.php

Both ' and ' are valid.

Well, it's cute that both are valid, but that's not the point...
get_html_translation_table is supposed to return "how php's functions translate it", not "any way which is valid".

And in that way, it _fails_ to do so.
Since the function html_entity_decode is only available as of php-4.3.0, anyone who has a similar function (based on the php-example on the documentpage!), finds it broken because of this.

In that sense it is, imho, a bug.
Quoting you're own documentation:
"get_html_translation_table --  Returns the translation table used by htmlspecialchars() and htmlentities()"

Not quite. When you have to write your own html_entity_decode(), you should cope with any forms of the numeric entity including hexadecimal style. It's not as simple as the snippet in the manual page.

Well, maybe so.

But I was refering to a function that tries to undo the changes of htmlspecialchars/htmlentities.

If htmlspecialchars changes ' to ' and you want to depend on get_html_translation_table to undo all changes, you expect it to return ' = ' instead of ' = ', since that's the change htmlspecialchars/htmlentities did aswell.
It didn't change it to '

If you really wanted to create a perfect entity-decoder, you'd indeed have to cope with all those &*; entities, including all the &#[0-9]{2,3};-like entities.

But for the simple "undo the htmlspecialchars"-like function that is not necessary.

And again, get_html_translation_table returns "how the htmlspecialchars/entities functions do it", not "all possible translations" or "just a valid version, maybe not what our own functions do", doesn't it? :)

To explain what I mean:
if you do 
echo html_entity_decode(htmlspecialchars("'", ENT_QUOTES));
you get ' back.

If you do:
function my_entity_decoder($string)
{
$trans = array_flip(get_html_translation_table(ENT_HTML_SPECIALCHARS, ENT_QUOTES));
$original = strtr($encoded, $trans);
}

echo my_entity_decoder(htmlspecialchars("'", ENT_QUOTES));
Where you trust the get_html_translation_table-function to return enough information to output ' again...

But if it all doesn't matter to you guys, why do the two change at all?
Why does the htmlspecialchars change it to ' why the get_html_translation_table claims it changes it to ' ??

Does the same in Windows PHP 4.3.4.

Fixed for PHP 5.3 and trunk.