StrConv() corrupting strings on Windows 2008 server

Question

I'm trying to migrate a web application from a windows 2008 vps to a dedicated windows 2008 vps ( both 32bit edition ). The problem i noticed ( and i've fighting with it for more than 24 hours as of now ) is that most of the text comes in a garbled form ( and this ruins all of the application ). the app works like this: (html+js) -> php -> vb6 dll -> reply.

Initially i thought it must be an issue with encoding and iis but i think this is not the case. I narrowed down the problem to this:

when functions of the .dll are called, the string parameters of the functions are first parsed and translated to a vb6 proper format:

Public Function getCorrectStringNoTrans(ByVal strWord As String) As String

    If strWord <> "" Then
        If isUTF8 = False Then
            strWord = Trim$(strWord)
        Else
            Open "c:\log.txt" For Output As #1
            Dim lol As Integer
            For lol = 1 To Len(strWord)
                Print #1, "bef:" & lol & ":" & Mid$(strWord, lol, 1) & ":" & AscW(Mid$(strWord, lol, 1)) & ":" & Asc(Mid$(strWord, lol, 1))
            Next lol

            strWord = StrConv(strWord, vbFromUnicode)

            For lol = 1 To Len(strWord)
                Print #1, "aft:" & lol & ":" & Mid$(strWord, lol, 1) & ":" & AscW(Mid$(strWord, lol, 1)) & ":" & Asc(Mid$(strWord, lol, 1))
            Next lol
            inData = strWord

            Dim c As New crc
            Dim lll As Long
            lll = UBound(inData)
            Print #1, c.CRC32(inData, lll)
            Close #1

            strWord = Trim$(ConvertUtf8BytesToString(inData))
            '''strWord = Replace$(Trim$(ConvertUtf8BytesToString(inData)), "\'", "'")
        End If
    End If

    getCorrectStringNoTrans = strWord

End Function

So in live machine 1 ( the current one, which works properly ) the output for a specific string is ( the last number is a crc check ) :

bef:1:Ξ:926:206
bef:2:Ί:906:186
bef:3:Ξ:926:206
bef:4:±:177:177
bef:5:Ξ:926:206
bef:6:»:187:187
bef:7:Ο:927:207
bef:8::140:140
bef:9:Ο:927:207
bef:10:‚:8218:130
aft:1:?:-17714:63
aft:2:?:-20018:63
aft:3:?:-17458:63
aft:4:?:-29489:63
aft:5:?:-32049:63
 1958248078 

but in the new machine the output is:

bef:1:Ξ:926:206
bef:2:Ί:906:186
bef:3:Ξ:926:206
bef:4:±:177:177
bef:5:Ξ:926:206
bef:6:»:187:187
bef:7:Ο:927:207
bef:8::140:140
bef:9:Ο:927:207
bef:10:‚:8218:130
aft:1:?:16191:63
aft:2:?:-20161:63
aft:3:?:-17601:63
aft:4:?:16191:63
aft:5:?:-32193:63
 523875602 

so ascw() shows that the string gets messed up after strconv() in the new machine i'm trying to migrate to.

i have triple-checked the settings in control panel -> regional and language options and all settings are the same in BOTH machines:

formats tab: current format: greek

location tab: current location: greece

administrative tab: language for non-unicode programs: greek

Then i thought that maybe it's an issue with SP2, because the correctly working machine was sp2, and this one i'm experimenting was sp1, so i updated but alas, no results.

I don't understand why strconv() doesn't work as it should. Maybe I'm missing something in the settings or installation. Any ideas would be highly appreciated.

p.s. it also works correctly in a windows xp development machine with iis5.

Answer 1

It seems that despite the fact that I have set language for non unicode programs to Greek ( codepage 1032 link: http://msdn.microsoft.com/en-us/goglobal/bb964664.aspx ), both:

Private Declare Function GetThreadLocale Lib "kernel32" () As Long
Private Declare Function GetUserDefaultLCID Lib "kernel32" () As Long

return 1033 ( US english ) in that specific box for UNKNOWN reasons. In the live box there is no such issue despite it being the same OS and settings

So what I did is change from :

strWord = StrConv(strWord, vbFromUnicode)

to

strWord = StrConv(strWord, vbFromUnicode, 1032)

and it works now.

Answer 2

You seem to be misunderstanding the encoding.If you're passed a string via COM, then it will already be unencoded and "raw" unicode (UCS-2) string. If you're being passed in UTF-8 you CAN NOT store it in a string until it's been decoded otherwise it will be corrupted. You must read non ANSI data as a byte array and decode as appropriate.

VB6 also has no native support for UTF-8. StrConv() with vbUnicode and vbFromUnicode are converting between byte arrays containing ANSI/MBCS data to VB strings in UCS-2.

In summary, assuming you are correctly being passed strings via COM, then the entire function can be replaced with Trim$().