Jeffrey Altman via RT wrote: > ma...@ntp.isc.org via RT wrote: >> Jeffrey Altman via RT wrote: >>> The leashw32.dll is filled with calls to MessageBox(). The assumption >>> was that if an error occurred within the krb5 library that it should be >>> displayed to the end user. The problem is that KFW is no used only for >>> interactive applications and even when it is the generation of >>> MessageBoxes from the library as opposed to reporting the errors to the >>> application and letting the application interact with the user is >>> (perhaps) a poor choice. >>> >>> Attached is a patch that disables all of the MessageBox calls within >>> leashw32. It permits USE_MESSAGE_BOX to be specified at compile time in >>> order to re-enable the MessageBox calls. >> It seems to me that a better solution would be to replace calls to >> MessageBox with a call to a different function and have that function >> decide how to deal with the message. Since this is a DLL it needs to >> decide at run-time whether or not it's interactive and what to do with >> the message. >> >> Danny > > Danny: > > I have to ask what benefit the extra work would provide? Simply > because the process happens to be running in an interactive environment > does not mean that a library should throw up arbitrary dialog boxes. > leashw32.dll was a support library for Leash32.exe and that ticket > manager relied on the library displaying the errors for help desk > support. Leash32.exe is no longer supported and cannot be compiled > with the latest Visual Studio. leashw32.dll is simply a support library > for gssapi32.dll and krb5_32.dll. In that role I do not believe there > is any reason to have it display dialogs. > > Jeffrey Altman
I didn't know the context. For this you may be right but as a general principle I encapsulate these things so that I have a generic function that can easily redirect the output based on the circumstances. Building for flexibility is harder than hardcoding. Danny -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean. _______________________________________________ kfwdev mailing list kfwdev@mit.edu http://mailman.mit.edu/mailman/listinfo/kfwdev