I have made suggested tests. There is indeed small bug in html entities handling in support for TextWrap pdf fields. When text is longer then place for printing, it was stripped and sometimes not decoded properly. This is the only problem found. Regarding display, encoded names are always handled properly, and edition does not change anything here. Fix for the pdf problem is here.
Small explanation to the db_escape function. We are not aware about backup file readability (they are created just for database restoring), our main concern is application security, and this is why db_escape works as is. When you will change the escaping code removing html entities encoding, you will make an application vulnerable to XSS attacks.
To prevent permanent XSS vunerability you have to prevent special html characters to be stored in database, or have to be encoded whenever it is displayed back to browser. We have chosen the former approach and implemented it in db_escape() function code (beside simple sql escaping).
The latter approach would require rewriting all the data displaying code in FA, which is just time wasting.
Regarding the lacking support for ISO-8859-2 in htmletities, this is old, never solved php problem, so we have to ommit is somehow. As all dangerous chars are included and have the same placement in both ISO-8859-1 and ISO-8859-2 sets, we can just fake encoding declaration here.
Janusz