Hi all!
I’ve just migrated from Etomite 0.6 to MODx (probably by doing it the "wrong" way ie. treating it as simply an upgrade from a previous version of MODx then correcting filenames that broke) and it all seems to be working quite happily ... but one very weird thing has happened. In various pages (not all but most), some (maybe 1% or 2%) of the spaces (as in " ", not " ") in the document have turned into unprintable characters (questionmark-in-a-diamond for Firefox, blank rectangle for IE)! Copying and pasting from either browser (running under WinXP Pro) into a hex editor reveals that the character appears to be 0xFDFF in unicode (little-endian - an "a" would be 0x6100). A non-Unicode (or at least I’m pretty sure it’s non-Unicode) client (gnu wget under WinXP cygwin in this case) results in the unprintable being 0xA0 instead.
There doesn’t seem to be much of a pattern to what spaces get replaced - some documents might get 3 spaces in a row replaced (that is, 3 spaces separated by words but with no other spaces between them), other documents might get them spread evenly. There also doesn’t seem to be much of a pattern to nearby characters and (as far as I can tell and this includes hidden linebreaks, tabs or whatever). As far as I can tell it’s only spaces that get replaced. If I edit the page in question using either the "none" editor (plain HTML) or TinyMCE, the unprintables still appear (within the editor itself). Of course, I could just replace them with spaces then save but it’s tedious, mucks up the date stamps and would get generally painfull.
I don’t recall having this problem in Etomite (well, OK thats not entirely true, I did have a problem at one point in circumstances that I don’t quite recall where I’d get one or two unprintables - not sure if they’re the same - pop up just after saving but that went away and I didn’t think much of it since). They don’t seem to be "growing" and a couple of recent documents I created (since migration) don’t seem to have this problem.
Does anyone have any ideas on where to look for the source and perhaps how to go about fixing this? Something tells me any batch fix will involve diving into the database somehow - something I’m not really that keen on doing!
Here are the other details:
PHP version: 4.4.2-1
OS version: Linux elgar 2.6.15.3, some distro of Debian (not entirely sure what, could probably find out of really necessary)
MySQL version: Ver 14.7 Distrib 4.1.15
Clients:
- Firefox 2/WinXP Pro (character shows as questionmark-in-a-diamond, pastes as 0xFDFF)
- IE 7/WinXP Pro (character shows as blank rectangle, pastes as 0xFDFF)
- IE6/Win2k Pro (character shows as question mark)
- wget/Cygwin/WinXP Pro (character shows as 0xA0)
Tnx in advance!
