Reload the search form
Supply GNU-grep style regular expressions for the header fields. ^, $ and | won't work. Spaces will be replaced with " \+" matching one or more spaces.The search is case-insensitive. Because of the search with grep and the format of database it is difficult to search for more than one keyword. This is not a messenger or SQL database. Please read the man page about GNU-grep if you don't know regular expressions.
For a start, if you type only one word in a search field, this word has to occur in the header belonging to the search field. All fields must match, empty fields match everything. A dot "." matches any character, an asterisk "*" allows 0 or more repetitions of the preceeding character. That means, ".*" matches everything. So, if you want to search for subjects which contain the words "problem" and "ghostscript" in this order but with anything between them, the correct query string is "problem.*ghostscript" in the subject field.
To give some internal information: We have made an index to our archive. Each line in this index corresponds to one article. This line consists of some fields (From, Subject, Message-ID, etc.) seperated by a special character. If the search is submitted we call a perl script. This script generates the regular expression that is handed to GNU grep to search the index. GNU grep is orders of magnitudes (no exaggeration) faster than anything else. Now here you see the problem: Searching for two keywords won't find all matches because "keyword1.*keyword2" won't match "keyword2.*keyword1", which one would like to find, too. So you better don't use multiple keywords. You also can see why $, ^ and | won't work...