Re: document class

If the user opens the display submenue by clicking on the specific
button, the following method will be called

void CMIAppView::OnBnClickedBtnDisplaySub()
{
    GetDocument()->StartSubMenu(this, CMIAppDoc::SMENUE_DISPLAY);
}

void CMIAppDoc::StartSubMenu(CWnd *wnd, UINT submenue)
{
    CWizardStartDialog dlgStartWiz(wnd);
    CDisplayPage DisplayPage;

    CStringList sListGetParams;

                 switch(submenue)
    {
    case SMENUE_NETWORK:
        break;

    case SMENUE_DISPLAY:

        dlgStartWiz.AddPage(&DisplayPage, CDisplayPage::IDD);
        dlgStartWiz.DoModal();

                                   /* the document class didn`t have
the latest information the user made in the
        dialog - so far: */

                                   //global variable DisplayInfo from
the CDisplay Class
 
DisplayInfo.GetParam(sListGetParams);

        //add these new information to the xml list
        XmlFile.UpdateXmlList(sListGetParams, submenue);

                                   break;

                      }//end of switch (of course there are more
cases... within the switch )

        //update all views
        UpdateAllViews(NULL, submenue, sListGetParams);
}

XML file represents document state;

CNetwork and CDisplay will have one global varialbe and the CXmlList
class will have a protected member installed in the document class,
which is called XmlFile (see the above method). Because I want to use
the doc/view modell - which means that the document will have all
information for the view and these information are stored in the xml
file.


nt managing the display?


de a pointer to the


bly even by


imply have a public


ccess the one-and-only


ct, I'd probably


ch as


hanism is completely


The CDisplayPage (dialog showing the display submenue) will have
access directly to the global CDisplay class variable to get the
parameters for all controls (checkboxen and so on) in the
OnInitDialog() class and before the dialog will be closed due to the
user clicked the OK button, the CDisplay class method SetParams() will
be called to verify the new settings and store them in the display
class.

I can`t see any advantage in extending the constructor to the
CDisplayPage (dialog) class to add the global variable of the CDisplay
class there instead of using the global CDisplay varialbe.


the dialog. There


..


Here`s the method I`m using to get the information from the CDisplay
class.

BOOL CDisplayPage::OnInitDialog()
{
    CWizardPage::OnInitDialog();

    CFont *font;
    font = DisplayInfo.GetStatFont();

    //set font to the controls
    m_EdtBright.SetFont(font);
    m_CheckDispOnOff.SetFont(font);
    m_StatBright.SetFont(font);

    //get params from CDisplay class
    CStringList slist;
    DisplayInfo.GetParam(slist);
    POSITION pos = slist.FindIndex(CDisplay::Brightness);

    //cedit control
    m_EdtBright.SetWindowText(slist.GetAt(pos));
    //checkbox control
    pos = slist.FindIndex(CDisplay::State);
    m_CheckDispOnOff.SetCheck(_wtoi(slist.GetAt(pos)));

    return TRUE;
}


 string lists and


ok but if I need to update several values (for exmaple in the network
dialog - ipaddr, subnetmask, gateway-addr, dhcp(enabled or not)), a
string lists make sense or not?


 coupling between


a is in XML. Nobody


f nobody looks at


ng on. If I


 which handled


siness I might


e)" which looked


 over a


t in my XML file


That means you will use Xpath to search for each specific item in the
xml file direct - or in a CMap, CstringArray where you have stored all
items from the xml file?
I mean I understand your approach, but for example the http soc class
will get some parameters from the document, I have to use a
PostMessage to the mainthread (because the http soc class uses not the
mainthread) and within such a message I`ve only one parameter wPARAM
to add informtion from the document because the other one is the
threadID of the http soc class.

//in the http soc class:
viswnd->PostMessage(UWM_GET_DATA_FROM_DOC,
(WPARAM)wPARAM, ::GetCurrentThreadId());

If I will ask for each value I`ll have to use 10x or 12x such
postmessages; I`m not sure but it sounds not really good to me that
this would be a good programming style.

viswnd->PostMessage(UWM_GET_IP_ADDR_FROM_DOC,
(WPARAM)wPARAM, ::GetCurrentThreadId());
viswnd->PostMessage(UWM_GET_SUBNET_MASK_ADDR_FROM_DOC,
(WPARAM)wPARAM, ::GetCurrentThreadId());
viswnd->PostMessage(UWM_GET_GATEWAY_ADDR_FROM_DOC,
(WPARAM)wPARAM, ::GetCurrentThreadId());

The advantage of using a CStringlist or CStringArray is that I will
pass these information by FormatMessage to the CString (in the http
socket class) which includes the html page (with %1%, %2% and so on).
So the list (getting back from the doc to the http soc class) includes
the information already in the right order for the specific http
page.

According to the CXML file class: at the beginning (in the
OnNewDocument method) I will call the xml class method to load the
specific xml file and load all the values / node values from the xml
file into a CStringList; after that I release the xml file.

The xml file looks like:

<myxmlfile>
   <display>
       <state checked=checked>1</state>
      <brightness disabled="">50</brightness>
      <contrast disabled="">50</contrast>
   </display>

</myxmlfile>

The display has three information in the xml file (brightness,
contrast and state which indicates if the display is on or off by a
checkbox); If the state is checked, the brightness and contrast is not
disabled (therefore I installed the attribute disabled=""). IF the
checkbox (state) would not be checked, brightness and contrast will
have the attribute in the xml file disabled="disabled".

So the xml file includes all information I need to know to get the
correct state and behaviour of each item regarding to the display... I
know if a ceditbox has the possibility to be disabled or not and so
on...

But I`m not really happy about using a CStringList for the xml file,
because it is a little bit complicated. I think there must be a quite
simpler solution to acchieve the same thing.

My CStringlist has the structure:

<menue display>
<!-- the next information are from the xml file -->
<first nodename> - <path to first nodename> - <value> - <attributes if
available> - <" " string which indicates the end of this node>
<second nodename> - <path to second nodename> - <value> - <attributes
if available> - <" " string which indicates the end of this node>
<empty string "">

<menue network>
....and so on

Is is possible to add more than one parameter to a CMap??? For
example: the node name is the key because its the only value which is
unique, the second parameter holds the path to the value in the xml
file (for storing this value later using xpath), the third parameter
holds the value and the forth parameter will hold any additional
attributes (not every item will have such attributes).

Then if the http socket class say: I need the ip addr (Postmessage
which includes the node name) -> I will take a look into the CMap
searching for the key (name of the node which holds the value and the
attributes) -> add all these information in the correct order into a
CStringlist or CStringArray and return this list to the http socket
class....

I know what you are thinking about that!? it`s really difficult and
not simple.... The key is that the http socket class as well as the
CNetworkDlg needs the value of e.g. the ip addr as well as the
behaviour of the ip addr (is it disabled or not) . The ip (which
represents a cedit field in the CNetworkDlg class) could be disabled
(if the user will use dhcp to get a ip addr) or the ip could be not
disabled (if the user will use a static ip configuration).

Which approach will you prefer to send the right information from the
xml file to the http socket class? I think the xml file class only
needs to know if a item could be disabled / checked / selected or not
- and the http socket class only wants to get some information to
display the page in the right way with all required information.

best regards
Hans