[VSERVER/V4RB] - Vserver on X.4.6, V4RB 2.3 Client Connection

[Russ Tyndall]

On 5/7/06 1:49 PM, "Russ Tyndall" <fitzbew at nc.rr.com> wrote:

> On 5/7/06 8:14 AM, "Russ Tyndall" <fitzbew at nc.rr.com> wrote:
> 
>> On 5/7/06 4:09 AM, "Ruslan Zasukhin" <sunshine at public.kherson.ua> wrote:
>> 
>>> On 5/7/06 2:38 AM, "Russ Tyndall" <fitzbew at nc.rr.com> wrote:
>>> 
>>>> This is what I am experiencing, although I do not know the cause.
>>>> 
>>>> I have an in-house app that's been used fairly heavily by two Mac OS X
>>>> machines, and is working pretty good.
>>>> 
>>>> Yesterday, I began to tweak the Win32 version of the client app.
>>>> 
>>>> Unfortunately, whenever I open the db from Win32, the db becomes unusable
>>>> by
>>>> the OS X clients.  When the OS X apps try to insert a record, all the
>>> 
>>> Do you mean by OPEN that AFTER WIN Client __connects__ to MAC Server,
>>>   do some work
>>>   disconnect
>>> 
>>> Then future MAC Clients connections get problems?
>>> 
>> 
>> Yes.  But the only "work" being done by the Win32 app is connecting and
>> opening the db and updating a single field in a User table that stores login
>> date/time.  After that, the Win32 app only creates a few ReadOnly vcursors
>> and later closes the connection.
>> 
>> Afterwards, when a Mac OS X (same RB 2006 app, just compiled for different
>> target) machine tried to insert a record, bad data is inserted into the
>> string and varchar fields.  Numeric data is stored properly.
>> 
>> The Vstudio diagnosis displays the problem described in the earlier excerpt
>> (included below).
>> 
>>>> string/varchar fields are not properly updated; numeric fields are updated
>>>> properly.  ThrowExceptions is set to True, but no exception occurs.  It's
>>>> just the fields are updated with "bad" data.
>>>> 
>>>> A diagnose from Vstudio reports this problem:
>>>> 
>>>> *****
>>>> 
>>>> ==========> Field: item_notes , type Text
>>>>     .RecID: 1440. Segment: 1441. already in use.
>>>> 
>>>> ******
>> 
> 
> Interesting addendum: When I compile the app as a stand-alone, and just move
> it over to the Win32 machine, it throws a vexception of "table does not
> exist".  Then when I remote debug the same project to the same Win32
> machine, I get the behavior that I described earlier (subsequent attempts by
> OS X clients to add records generate messed up records).
> 
> I'd like to also clarify that problems are found in the DB by Diagnose after
> the Win32 client opens and closes it, and *before* any OS X clients touch
> it.
> 
> 2.3b10 is also installed on the Win32 machine, I assume. The latest
> modification date in the vcomponents file is April 9.
> 
> The RB IDE has the V4RB carbon plugin installed.

Working just within the RB Debugger and Vstudio:

If I export just the structure of the db, and import it into a new db (mode
4, but the original db is mode 1) AND change a Text field in a table to
VarChar, I can get cross-platform to work a *few* times, but eventually the
Mac OS X client begins inserting what looks like gibberish (or some encoding
I do not recognize).

Before I spend more hours trying to narrow this down, may I ask:

Is anyone using Vserver 2.3 on X.4.6 with RB 2006 clients from XP
successfully? Updating tables with a mixture of varchar and text fields?  I
guess it would give me peace of mind to know that this *can* work, and that
I am suffering from some local issue.

The OS X RB clients (virtually the same app) continue to happily work with
Vserver, no problems whatsoever (well, nothing like this Win32
show-stopper).

-- 
Russ Tyndall
Wake Forest, NC