C / C++ / MFC

9-Jul-03 11:15

Why would this:

fwrite(dataPtr,sizeof(unsigned char),(sizeof(unsigned char)*datalength),filePtr);

ALWAYS return 0 byte written Confused | :confused:

I can open the file ok using

filePtr = fopen (DataFile,"wb");

but whenever I try to write the data down it fails.

Anyideas?

Thanks!

Led9-Jul-03 11:23

9-Jul-03 11:23

are you sure your dataPtr is not NULL ?
(or 0 - whichever your prefer Smile | :)

)

will13839-Jul-03 11:26

9-Jul-03 11:26

lol

I just checked that and it seems my data pointer is not pointing to the right place. LOL. fix that and then try again.

Thanks. I might be back. $Shucks | :-\$

will13839-Jul-03 11:32

9-Jul-03 11:32

ok pointers are all correct now, data is where I need it to be. But still, no data being written when calling the fwrite.

Unsure | :~

Neville Franks9-Jul-03 12:14

9-Jul-03 12:14

Did the file open succeed? If you look at the data buffer in the debugger is it what you expect? What does errno tell you? Why aren't you checking what fopen() and fwrite() return.

Neville Franks, Author of ED for Windows. Free Trial at www.getsoft.com

Mustafa Demirhan9-Jul-03 12:54

Mustafa Demirhan

9-Jul-03 12:54

Firstly, your code should be like this:

fwrite(dataPtr, sizeof(unsigned char), datalength, filePtr);

It does not make any difference in case of a char, but if you use integers for example, your code will try to write 4x4xdatalength = 16xdatalength bytes of data, instead of writing 4xdatalength bytes.

Anyway, apart from that be sure to check the datalength. Check out whether it is 0 or not.

If it still does not work, add the following line after writing to the file:

fflush (filePtr);

Mustafa Demirhan
http://www.macroangel.com
Sonork ID 100.9935:zoltrix

<nobr>They say I'm lazy but it takes all my time

Finally!

will138310-Jul-03 1:14

10-Jul-03 1:14

Ok,

I did finally find it. Turned out that my global datasize number was being clobbered all the time, aka ending in a data size of 0 to be written.

Thanks for all of your help! I REALLY appreciate it!

Woo HOO!!

Dan

Confused in abby

PJ Arends9-Jul-03 11:03

PJ Arends

9-Jul-03 11:03

Dear Forum,

Ok, this is really starting to bug me. Anybody have any idea why the 'TableName' is not being added to the string 'str', even though the 'TableName' seems to be valid as the watch window shows the correct value and the TRACE call does manage to successfully out put it?

for (std::vector<DFields *>::iterator it = pConfig->Samples.begin(); it != pConfig->Samples.end(); ++it)
{
    CString str = _T("");
    if (pConfig->IncludeComputer)
    {
        str += (*it)->ComputerName;
        str += _T(" : ");
    }
 
    if (pConfig->IncludeTable)
    {
        TRACE("*%s*\n", (*it)->TableName); // TableName successfully TRACEd
        str += (*it)->TableName;           // Why TableName not added to string????
        str += _T(" : ");
    }
 
    str += (*it)->SampleName;
 
    m_ListSelected.AddString(str);
}

After this function, 'str' is (sans quotes): "Barn 1 Feed : : Total Daily Feed"

and the TRACE output is: "*743 Daily*".

str should be: "Barn 1 Feed : 743 Daily : Total Daily Feed"

I have tried to do a rebuild all, exit dev studio and then rebuild all, even rebooted my computer. But still no luck. All I know is that I will probably see what I am doing wrong as soon as I hit the submit button Unsure | :~

[edit]
ComputerName, Tablename, and SampleName are all CStrings.
[/edit]

Signed
Confused in abby

Sonork 100.11743 Chicken Little

"You're obviously a superstar." - Christian Graus about me - 12 Feb '03

Within you lies the power for good - Use it!

Re: Confused in abby

Led9-Jul-03 11:26

9-Jul-03 11:26

I don't have a clue,
but have you tried casting the TableName to f.e. (char *) or something alike..?

I wouldn't be amazed if it was some sort of type-confusion..

Re: Confused in abby

PJ Arends9-Jul-03 11:32

PJ Arends

9-Jul-03 11:32

hmmm. That worked. wierd Confused | :confused:

Thanks a lot Big Grin | :-D

Sonork 100.11743 Chicken Little

"You're obviously a superstar." - Christian Graus about me - 12 Feb '03

Within you lies the power for good - Use it!

Re: Confused in abby

Neville Franks9-Jul-03 12:11

How to keep rendering during user-input

9-Jul-03 12:11

Its because your mixing crappy old MFC CString and lovely new STL, you naughty boy you. Big Grin | :-D

Neville Franks, Author of ED for Windows. Free Trial at www.getsoft.com

Project Converter

308holes9-Jul-03 10:56

308holes

9-Jul-03 10:56

Hello i was told to look here for something that would convert a VC++ 7.10 Project to a VC++ 7.0 project can anyone help me ??? Confused | :confused:

Re: Project Converter

Mustafa Demirhan9-Jul-03 12:44

Mustafa Demirhan

9-Jul-03 12:44

Here it is[^] Wink | ;)

Mustafa Demirhan
http://www.macroangel.com
Sonork ID 100.9935:zoltrix

<nobr>They say I'm lazy but it takes all my time

Led9-Jul-03 10:38

9-Jul-03 10:38

Hi,
I'll try to explain my problem in as few words as possible so I won't scare anyone off with long posts Smile | :)

Mission :
I'm working on a hobby-project to render graphics that change with user-input.

Problem :
GfxObjects can be deleted/changed, but *during* rendering that's a big NONO as you'll probably understand Smile | :)

Possible solutions :
- Multithreading.
Tried it, it's a very big pain in the arse.
GfxObjects could be plugins with their own imported edit-dialogs and I don't want other developers to have the burden of doing locking / unlocking everywhere (and give'em a perfect opportunity to crash my app).

- Singlethreaded :
OnIdle is fine, it's just not enough.
F.e. when the mouse enters the menu, idling stops.
Same for window-resizing etc.

The ideal way would be :
{
ProcessUserInput( [for a maximum time of 1/100th sec.] );
RenderFrame();
}

I hope some of you guys have suggestions on how to tackle this Smile | :)

-

Lennart Denninger / GameCoder / Guerilla Games

Neville Franks9-Jul-03 12:17

9-Jul-03 12:17

Well like it or not I think you need to use the multithreading approach to get the result you want. One interuptable thread for rendering, and UI back in the main app thread.

Neville Franks, Author of ED for Windows. Free Trial at www.getsoft.com

Led9-Jul-03 12:51

9-Jul-03 12:51

I know...

But it would be nice to have a way to lock the object-system before all the ui-handling and unlock it afterwards.
Just like locking it before rendering and unlocking it after rendering.

Locking / unlocking for every single change sucks,
it means that for someone to write a plugin with an edit-dialog the code would look like this :

CEditDlg::OnVar1Changed()
{
LockObjectSystem();

// change var here
mObject->SetVar("General.SplineSegments", 10);

UnlockObjectSystem();
}

I'd prefer something like

{
LockObjectSystem();
UpdateALLDialogsAndSettings();
UnlockObjectSystem();
}

'cause in this case the people creating plugins wouldn't need to know about all the locking-stuff, and more important - they wouldn't be able to send my app down in flames by simply forgetting to lock/unlock.

Neville Franks9-Jul-03 13:08

9-Jul-03 13:08

Why can't you collect the info they enter in the plugin and then apply it when the plugin finishes. This puts you in total control and you can handle any locking required.

Can't say I comprehend precisely what you are getting at with locking, but I assume you only want one user/thread/plugin to be able to change objects at a time.

Neville Franks, Author of ED for Windows. Free Trial at www.getsoft.com

Led9-Jul-03 13:40

9-Jul-03 13:40

A message-based system like you propose has crossed my mind (and I even started implementing it),
but it seems like an overly complex solution to a very simple problem Frown | :(

Then I'd have create a message-system that handles a lot of different (custom) messages,
lock/unlock a message-queue etc....

If only there was a way to
1st) update the UI and thus the gfx-object parameter changes
2nd) render the frame
where the rendering would be virtually uninterrupted, and the UI would be allowed to have some lag...

The plugins don't "finish" by the way;
it's a normal app with an edit-window for the configuration of the selected effect (selected from a listcontrol in another window).

So the render-thread is rendering non-stop,
and meanwhile the user can change effect-properties in the UI.
(Think of a winamp-plugin with a realtime edit-window)

Still, right now I'm thinking of going back to the multithreaded system..
But then I still want to find a way to make sure the plugins can't crash the app.
Which means taking the locking/unlocking out of their hands - but I still have to find an elegant way to do that :-/

Led9-Jul-03 13:51

9-Jul-03 13:51

Forgot to mention this :

The (plugin)edit-dialog also has to display the current values for all the effect-parameters.

Which would mean for a message-based system like you proposed (gathering all entered data and processign it later),
that the message-system has to be bidirectional.
(Since you're not only inputting new values, but also *reading* them from the object)

And that's a whole different, (imho way *too*) complex ballgame.

Frown | :-(

Ps.
I don't want to disrespect you or anything,
(I really value your input on my problem !!)
but your signature looks really
"Hey! look at me! I'm a professional developer!"
flashy kind of thing Smile | :)

Isn't it a bit too much..? Wink | ;-)

Neville Franks9-Jul-03 15:35

9-Jul-03 15:35

Is it feasible to have two instances of the "object system". One is used by the renderer and the other by the UI components. Whenever the renderer has finished etc. it asks the UI object if it has changed, and if so it copies whatever it has to, to get in sync. A lock would only be necessary during this sync process.

re. my signature, you should have seen it a year or so back. Wink | ;)

I'm open to suggestions to tone it down further?

Neville Franks, Author of ED for Windows. Free Trial at www.getsoft.com