I've been customising Ben Poole's dominoWiki to suit our internal needs (adding things like page renaming and disambiguation for example).   So far so good.

As users add more varied content and experiment with the syntax we discover the odd strange anomaly.  Here's an example:


The user adds some content to a wiki page like this:

ABCDEFGHIJKLMNOPQRSTUVWXYZ
 abcdefghijklmnopqrstuvwxyz
 1234567890
 ,./;'<>?:@[]{}!"£$%^&*()_-+=
 [
 ]
[ ]
A [ ]

and when saved the render output looks like this:


ABCDEFGHIJKLMNOPQRSTUVWXYZ
 abcdefghijklmnopqrstuvwxyz
 1234567890
 ,./;'<>?:@[]{}!"£$%^&*()_-+=
 
 ]
[ ]
A [ ]

If you look carefully you'll see the " [" after the "-+="  and newline disapears!

You could test this yourself by creating a form in a database with a single rich text field called "Body".  Create a WebQueryOpen agent with the following code & have the form run the agent when the document is opened on the web.

Sub Initialize
    
    Dim session As New NotesSession
    Dim content As String
    Dim doc As NotesDocument
    Dim niBody As NotesItem
    
    Set doc = session.DocumentContext
    Set niBody = doc.ReplaceItemValue("Body","")
    
    content="    ABCDEFGHIJKLMNOPQRSTUVWXYZ"
    content= content + Chr$(10) + "    abcdefghijklmnopqrstuvwxyz"
    content= content + Chr$(10) + "  1234567890"
    content = content + Chr$(10) + {  ,./;'<>?:@[]}+ "{}" +{!"£$%^&*()_-+=}
    content= content + Chr$(10) + "  *[*"
    content= content + Chr$(10) + "  ]"
    content= content + Chr$(10) + "["
    content= content + Chr$(10) + "]"
    content= content + Chr$(10) + "[  ]"
    content= content + Chr$(10) + "A [ ]"
    
    content = massageChunk(content, True)
    
    Set niBody = doc.ReplaceItemValue("Body",content)
    
End Sub


Private Function massageChunk(txt As String, noParaTag As Boolean) As String
   
    If noParaTag Then
        massageChunk = Replace(txt, Chr(10) & Chr(10), "<br />")
        massageChunk = Replace(massageChunk, Chr(10), "<br />")
    Else
        massageChunk = Replace(txt, Chr(10) & Chr(10), "<p>")
        massageChunk = Replace(massageChunk, Chr(10), "<br />")       
    End If
   
    massageChunk = Replace(txt, Chr(10) & Chr(10), "<p>")
    massageChunk = Replace(massageChunk, Chr(10), "<br>")       
   
        '** special case where </h3><br> adds too much line spacing
    massageChunk = Replace(massageChunk, "</h3><br>", "</h3>")
    massageChunk = Replace(massageChunk, "</h4><br>", "</h4>")
    massageChunk = Replace(massageChunk, "<br><li>", "<li>")
    massageChunk = Replace(massageChunk, "</li><br>", "</li>")
    massageChunk = Replace(massageChunk, "</ol><br>", "</ol>")
    massageChunk = Replace(massageChunk, "</ul><br>", "</ul>")
    massageChunk = Replace(massageChunk, "<br><dl>", "<dl>")
    massageChunk = Replace(massageChunk, "<br><dt>", "<dt>")
    massageChunk = Replace(massageChunk, "<br><dd>", "<dd>")
    massageChunk = Replace(massageChunk, "</dl><br>", "</dl>")
    massageChunk = Replace(massageChunk, "</dt><br>", "</dt>")
    massageChunk = Replace(massageChunk, "</dd><br>", "</dd>")   
   
   
End Function

When you open the form on the web you should see the problem.  Now edit the agent so that line
content= content + Chr$(10) + "  [" becomes content= content + Chr$(10) + "  *[*" and you'll see that the *[* appears in the ouput where the "[" character had been dropped.

I'm guessing that it has something to do with Domino pre-processing content for the web and thinking that opening bracket denotes pass through HTML.   That said why isn't the corresponding "]" also dropped along with subsequent "[" "]" pairings? 

It's clear to me (through testing) that the Replace function isn't at fault.

Does this make any sense?  Am I missing something?

Jason

(p.s. I have another anomaly up my sleeve for later today if I get time )