>>> >>> >>> http://zhadum.east.sun.com/export/ws/amaguire/nwam1-fixes/webrev/ >>> >>> >> General: >> >> [...] >> >> Also, one thing that I've noticed is that when NULL handles are >> passed to the library (say with nwam_*_read()) and the read fails >> because the entity does not exist, the handle is not NULL anymore. If >> a NULL handle is passed, should it be set back to NULL when not >> returning NWAM_SUCCESS. (nwamadm uses NULL handles after the code >> review and every time _read(), _create(), _copy(), etc fails, handles >> have to be set to NULL). > That's annoying - I'll fix this. If its too much, I can take care of this. I just went through adding handle=NULL in nwamadm, so I am familiar. In the library before "return (...)", "if (*handle == NULL) handle = NULL". >> >> nwamd/ncu.c:596 - what's the reason behind setting the user_enabled >> to TRUE when the enabled property cannot be read? >> > It's for prioritized NCUs that don't have an enabled property - > user_enabled is probably a misleading name though. There the "enabled" member in nwamd_ncu_t (and also user_enabled). "enabled" doesn't seem to be used anywhere. It appears that we want the variable to be called "enabled", which is set to FALSE for manual NCUs that are disabled and to TRUE for prioritized NCUs and manual NCUs that are enabled. Also, a comment on ncu.c:596 will make things clear.
Anurag
