|
I think that code snippets for the dec. of the structure, the writing of records, and the reading of records. Packing/alignment and how you "changed" members of the structure to other structures may be part of the problem.
Oh, and ReadFile(...) and WriteFile(...) are the Win32 APIs for reading and writing files. fread(...) and fwrite(...) are C-RTL functions.
Peace!
-=- James
If you think it costs a lot to do it right, just wait until you find out how much it costs to do it wrong!
Avoid driving a vehicle taller than you and remember that Professional Driver on Closed Course does not mean your Dumb Ass on a Public Road!
DeleteFXPFiles & CheckFavorites (Please rate this post!)
|
|
|
|
|
thanks james, But can i write records into a file using ReadFile() and WriteFile() and also i have another question please clear the doubt. if i am reading the file from a secondary thread and at that moment the primary thread tries to update it then, will it generate any error letting my second thread crash ? i cant let this happen as the secondry thread runs in the background. how can i achive the synchronization using WIN32 code....
kss
|
|
|
|
|
No, I do not think you will have any errors unless you manually handle locking sections of the file. You should synchronize access to the file HANDLE by wrapping calls to your read/write functions inside of a critical section (if the threads are in the same process) or a named object (event, mutex, etc.) if they are in separate processes.
Look up the Sync objects and their related functions (SetEvent, WaitForSingleObject, EnterCriticalSection, etc.) in the MSDN help
Peace!
-=- James
If you think it costs a lot to do it right, just wait until you find out how much it costs to do it wrong!
Avoid driving a vehicle taller than you and remember that Professional Driver on Closed Course does not mean your Dumb Ass on a Public Road!
DeleteFXPFiles & CheckFavorites (Please rate this post!)
|
|
|
|
|
fwrite and fread are OK.
BTW, why not use a database instead of managing records manually?
|
|
|
|
|
no I cannot because i am writing a plugin kind of thing. so i can't use any database.my plugin has to read this file every 6 seconds to check for any new updates hey BTW will it create a problem while i am reading from the file and another therad updates this particular file. will the reading thread crash ?..
kss
|
|
|
|
|
fearless stallion wrote: BTW will it create a problem while i am reading from the file and another therad updates this particular file. will the reading thread crash ?..
No, but the data you'll read will be corrupt. Of course, if you manage sharing properly the reading thread will fail to open the file instead.
|
|
|
|
|
|
I do stay away from design meetings 
Allow me to quote a part of the original message:
"the Structure i am writing into file contains char fn[100];char sn[1000];char dt[20];char tm[20]; int ET;"
Now this is big enough for a single record. Since records are used, it's more than probable that these will be searched or accessed randomly. Fine with me if there are a couple of records, but what if there are much more?
Besides, database is just an idea, not an imperative. I'm not the guy's boss, so I'm not imposing this design alternative as a "must", only as a way he might have failed to consider.
Finally, who mentioned a full-blown db? I'm not sugesting a db server and such. A small and handy MSAccess db can do the job more effectively. Especially when there are two threads accessing the file simultaneously (as is the case - see previous messages).
Sure you understand all this as well as I do so there's no reason why I should be subject to such criticism is there?
|
|
|
|
|
HAND wrote: Sure you understand all this as well as I do so there's no reason why I should be subject to such criticism is there?
Yes, of course. You will note, of course, the amount of simileys used in the message, and even on the icon for the post itself. They are there for a Reason.
...But since you asked...
Those records are small, and there appears to only be one "entity". With a record size of ~1126 bytes, even 10000 of them can be managed in memory rather easily (These days, 11MB is a trivial amount of data to manage), and can be saved and loaded from storage quite easily as well. No need for additional requirements, even to bring DAO or a similar technology into the picture for something as small and handy as an MDB file. I would question anyone that suggests any kind of external database implementation (ranging from SQLite, to MS Access, to a full-blown C/S RDBMS implementation) for a single entity.
IME, experienced VC++ developers do not consider Access to be a "database". So when someone mentions the word "database", we do not envision an MDB file, but a true, mature, capable RDBMS implementation. If you would consider a flat-file storage implementation a database, then that is what the OP has already, and there is no need to mention "database" at all.
Peace!
-=- James
If you think it costs a lot to do it right, just wait until you find out how much it costs to do it wrong!
Avoid driving a vehicle taller than you and remember that Professional Driver on Closed Course does not mean your Dumb Ass on a Public Road!
DeleteFXPFiles & CheckFavorites (Please rate this post!)
|
|
|
|
|
Oh I see. It wasn't obvious to me the entity is single, hence my suggestion. If it is, I DID sound rather inappropriate not to say stupid 
Oh and - why not call MSAccess *.mdb file a database? Access supports SQL standards (mostly) and isn't necessarily flat as such (although here it's flat as we're talking about a single table, of course). The only problem with MSAccess is the oh-so-slow Jet Engine.
Peace!
|
|
|
|
|
Again, I have the problem of associating "database" with a full-fledged RDBMS system, not one that suffers limitations that one would not expect.
Off of the top of my head, I would consider things like:
o the lack of true user-level security (all you really need is Permission access to the underlying physical file), excluding DB passwords
o the lack of complete join support (no OUTER and FULL join, last I knew; not sure about the new Access DB engine, which is NOT going to be JET-based)
o the lack of true "server side" processing - since the DB is just a file, and the data processing components are on the client-side, if you have the file remote and run a query, data is sent over the wire and then processed by the client.
IMHO, MS Access, like Java, has a place. In the case of MS Access, it allows non-tech types to build easy form-based data storage and processing applications. But since I deal in large applications and systems, MS Access generally does not have a place in there. I have created and used JET databases in applications in the past, and I remember myself always wanting something more from them. These days, I would not suggess MS Access off-hand - at least, not without much more research into the needs of the project in question.
Peace!
-=- James
If you think it costs a lot to do it right, just wait until you find out how much it costs to do it wrong!
Avoid driving a vehicle taller than you and remember that Professional Driver on Closed Course does not mean your Dumb Ass on a Public Road!
DeleteFXPFiles & CheckFavorites (Please rate this post!)
|
|
|
|
|
You arose quite a discussion! Still, no answer to your original question I believe?
As far as I can see, your code should work just fine, but are you sure that, after changing your structure, you deleted the original file? Could it possibly be that you are reading information that was written with the old structure? In that case, anything could happen!
The problem here is that you simply write a block of data, assuming that it must be what you expect it to be. If you need to be more sure, also after changing the data's structure, you would need to write individual data elements, with some kind of indication as to thei meaning (something like ASN.1, but you could of course write your own simpler version).
If this is no problem to you, then what you are doing now should work. If you simultaneously write from one thread (or application) and read from the other, then the data you read might be corrupted! (imagine one thread being halfway writing a tm structure, while the other is reading it). The best way around this is using simple file-locking.
If you use the CFile method, this would be "CFile::LockRange ()" for example.
If you use the FILE * approach, you will have to use fsopen to open the file, using for example _SH_DENYWR to avoid someone else writing to the file.
Success,
William
|
|
|
|
|
Thanks willam,
Let me put it a bit more clearly,
The primary thread of my application or the main thread launches the second worker thread when it is initialized. the second thread runs in the background and after every 6 seconds looks for a file say "abcd.txt" where records are written. now records to this file are written from the primary thread. and this writing can take place at any moment the enduser wishes. say at a particular moment the secondary thread is reading the file and the enduser writes to it at that particular moment than the data read from the file will be corrupt. that is what my problem is. i don't want to deny the end user ,rather i would like to keep him waiting for a second or two and then once the read is complete, i would like to write into the file. and give him a success message.
The critical section point that james made above sounds nice. but if you have any other suggestion then please let me know.
One more question , is it a good idea, reading from a file every 6 seconds?
kss
|
|
|
|
|
My best suggestion would be to use locking. I don't know how big your file is going to be, but I assume that this will not be enormous? In that case, reading will only take some milliseconds, so your user will not even notice that he has been waiting for the read to complete.
On the other hand, locking is a two-way process: also your reading thread will have to wait for the write to complete. However, this too should take just milliseconds. Besindes, you could fine-tune all of this, by just locking the record that you are working with (the LockRange instruction allows you to specify a starting point and a size of the range you want to lock). This makes the chance of having to wait even smaller.
Success,
William
|
|
|
|
|
The file can contain up to 1000 such records. That makes it of 1.1 mb or 1.2 mb . I was looking at the MSDN and saw the CMemFile class can i use it in a Win32 Project ? is there any equivalent of LockRange win32 APIs ? or Do i have to use the CFile::LockRange ?
kss
|
|
|
|
|
In general, you do not delete items from an ImageList. An Imagelist associated with a ListView control usually does not contain one icon for each item, usually all of the images use one of a few different icons.
Think of it like Windows Explorer - most normal folders share the same image. Just because you remove the last folder in the view does not mean that the folder image gets deleted as well.
If you are trying to have a specific, unique icon for each ListView item, and you have lots of items, an ImageList is likely not the best way to go - I would look into using CustomDraw or virtual mode to draw the icons as needed.
Peace!
-=- James
If you think it costs a lot to do it right, just wait until you find out how much it costs to do it wrong!
Avoid driving a vehicle taller than you and remember that Professional Driver on Closed Course does not mean your Dumb Ass on a Public Road!
DeleteFXPFiles & CheckFavorites (Please rate this post!)
|
|
|
|
|
Hello !
I would like to print on the console (in a console application) the status of a calculation which runs in a separate thread. I would like in fact to put the evolution of the calculation (0 - 100 %). The problem is that I don't know if this is possible to write data always at the same position in the console (otherwise, this will look very ugly).
Thanks for the suggestions.
|
|
|
|
|
try gotoxy() or something
divin
|
|
|
|
|
|
there is a function to place the cursor at a specific (x ; y) couple -not GotoXY() which is borland specific), but i don't remember which.
what you can do is also to use the '\b' (backspace) character to go back on the characters you've just printed out...
printf("12.37");
printf("\b\b\b\b\b"
"14.17");
this will actually print the last 14.17 only...
you also have the possibility to use the '\r' line feed character to go back to the beginning of a same line...
TOXCCT >>> GEII power
[toxcct][VisualCalc 2.24][3.0 soon...]
|
|
|
|
|
Easier; uise a carriage-return:
while (Loop)
{
... do something ....
printf ("12.12\r");
}
printf ("\n");
Don't forget the last line-feed, since otherwise the prompt will appear within this line after your process terminates.
Success,
William
|
|
|
|
|
yeah, i know, this...
but sometimes, you don't want to overwrite a whole line but only some characters...
TOXCCT >>> GEII power
[toxcct][VisualCalc 2.24][3.0 soon...]
|
|
|
|
|
oh, i found it !
see if using this don't help you :
header : <wincon.h>
method : SetConsoleCursorPosition()
TOXCCT >>> GEII power
[toxcct][VisualCalc 2.24][3.0 soon...]
|
|
|
|
|
Thanks to everybody !
Yes, of course I could have used the back character. I think this is the easiest solution.
|
|
|
|
|
I have a question about the printing process :
In my program, i want to make a button to print directly the window, without any dialog box. But when i call the OnFilePrint() function in my class CmyView, but it doesn't work.
How do I bypass this dialog box message ?
I have found this but it also doesn't work, the box is here again...
http://www.microsoft.com/msj/0197/c/c0197.aspx
-- modified at 5:55 Thursday 23rd February, 2006
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
Yikes! 
