Hi, > testwmtx.prg may not work because Harbour internal can call Windows API and > setting "Last Error". > [I think in MT mode, with hbvmmt, does not work]. > " > Some functions set the last error code to 0 on success and others do not. > You should call the GetLastError function immediately when a function's > return. > "
Quite strange behavior :/ [ what if another thread kicks in between function call and GetLastError() call...? ] > This means that wapi_GetLastError() can not be a wrapper 1:1 to Windows API > GetLastError(). > > Two solutions .- > One is to maintain a dynamic variable, thread safe, for each thread/process > to store last error and return this value in wapi_GetLastError(). > And another solution is to return last error as the last parameter of each > wrapper function to Windows API that can set and remove wapi_GetLastError() > because returns values can not be true. > > IMHO to Harbour this latter solution is probably the most easiest and > effective in terms of performance. > For example, with this solution testwmtx.prg would be like this .- > > What do you think about this, comments? I'd vote for first option, as it's the only one that allows to keep full Windows API compatibility for WAPI_* functions. Since this is the main point for all WAPI_* functions (and the whole idea), it's essential to stick to this rule. Otherwise we will end up resorting to workarounds (altering calling method) when return value is also some meaningful value and we want to return that plus lasterror. Brgds, Viktor _______________________________________________ Harbour mailing list (attachment size limit: 40KB) Harbour@harbour-project.org http://lists.harbour-project.org/mailman/listinfo/harbour
