Please see this answer[^] Regards, mav -- Black holes are the places where God divided by 0...

:) It could be but ... CryptAcquireContext() does not raise this error, it succeeded. It's CryptDecrypt() which fails. Naturally this function tries (would?) to decrypt data using the AT_KEYEXCHANGE key obtained with CryptGetUserKey() from the context acquired from the smart card. According to MSDN: "The CryptGetUserKey function retrieves a handle of one of a user's two public/private key pairs..." If I'm not wrong CryptoAPI CryptEncrypt()/CryptDecrypt() should use the public key to encrypt and the private key to decrypt, but decrypting with the private key from the smart card fails. Changhing the key type to AT_SIGNING when retrieving the key pair with CryptGetUserKey() has no effects (and it has not sense for me. I don't want to sign a piece of data with the private key. I want decrypt a piece of data with it ).

I'm using VC++ without managed code. However it seems that DPAPI could solve this problem, either using a machine-store or a user store. For the latter, a point is not clear for me: for user-store it seems DPAPI use the windows user logon credentials to generate the master and session keys. The CryptProtectData and CryptUnprotectData functions have the chance to show a dialog to allow the user to insert a password. Is this only necessary if the windows user has no password set (may be a standalone machine) ?

This could really be the answer ! But I should exhume the code wherever it is now, after three years. However this seems an useful clue. Thanks. :)

Take a look at where and how you create an instance of the class. The probable cause is that you have created an instance somewhere with the keyword new, and you are failing to call delete. If that class has other classes as members, their destructors will in turn fail to be called.

Nyarlatotep wrote: Do I have to use exception handling ? yes use try...catch nave

I'm not sure what effect all of those other functions have on the date value, but I tried the following and got the correct output both times regardless of the value of tm_isdst: struct tm now; now.tm_sec = 0; /* seconds after the minute - [0,59] */ now.tm_min = 0; /* minutes after the hour - [0,59] */ now.tm_hour = 15; /* hours since midnight - [0,23] */ now.tm_mday = 1; /* day of the month - [1,31] */ now.tm_mon = 5; /* months since January - [0,11] */ now.tm_year = 106; /* years since 1900 */ now.tm_wday = 5; /* days since Sunday - [0,6] */ now.tm_isdst = 1; printf("%s", asctime(&now)); time_t t2 = mktime(&now); printf("%s", ctime(&t2)); I guess I'm just not understanding your problem well enough. "Money talks. When my money starts to talk, I get a bill to shut it up." - Frank "Judge not by the eye but by the heart." - Native American Proverb

Forgive me :) I'll post in the c++ thread ...

Nyarlatotep wrote: The question is why GetTextMetrics hangs. Never happened before !!! This may due to following. -You have taken the device context of the printer where lots of jobs may pending -When you get the device context of printer and textout the text the device context is in use while jobs may pending or your process is not scheduled. -and you are trying to get/set the information of the device context which may denied as exclusive access to device context:~ Hope you got the issue. Knock out 't' from can't, You can if you think you can :cool: