I have the impression that this is an issue with the underlying ADOdb library - need to test.

It seems that the recordset's fields' case is not properly set to lowercase as it should:

In check_database_inc.php, the following query is executed: 'SHOW TABLE STATUS'.
The first fetched record - via db_fetch_array() - has lowercase field names, as expected (e.g. 'name').
Subsequent records have the first char of the field in uppercase ('Name')

Just want to be sure that you did notice also this special treatment in check_database_inc.php

switch( $g_db_type ) {

case 'mysql':

$t_field_name      = 'Name';

$t_field_comment   = 'Comment';

$t_field_collation = 'Collation';

$t_field_field     = 'Field';

$t_field_type      = 'Type';

break;

case 'mysqli':

$t_field_name      = 'name';

$t_field_comment   = 'comment';

$t_field_collation = 'collation';

$t_field_field     = 'field';

$t_field_type      = 'type';

break;

}

Thanks, I know about that one. The problem is that these variables are used as index for the rows being retrieved further down,

$t_result = db_query( 'SHOW TABLE STATUS' );

while( $t_row = db_fetch_array( $t_result ) ) {

and db_fetch_array() is returning the first row like this

Array

(

[name] => mantis_api_token_table

[engine] => InnoDB

[version] => 10

...

and the subsequent ones like this

Array

(

[Name] => mantis_bug_file_table

[Engine] => MyISAM

[Version] => 10

...

It looks like the ADOdb ADORecordSet::MoveNext() method is not properly handling the field names' case in associative fetch mode.

This issue also exists with ADOdb 5.19. Further research is needed, but I don't have time to do it now.

Confirmed ADOdb issue https://github.com/ADOdb/ADOdb/issues/167.

The problem is not as critical as I thought initially, because in our case it only affects queries on the MySQL information_schema. Mantis schema queries are not affected, because all our column names are in lowercase.

ADOdb 5.20.2 which includes the required fix is expected to be released next week.