Hello all, during my journey through the openser code is noticed a few places for improvement.. The same issues exists with some of our internal modules.
There is a huge duplication of effort regarding utility functions for data type conversion or data handling in general. Many modules defines their own, slightly different functions. Only for str and char* copying there existing at the moment at least five different functions in the code: str_dup, str_duplicate, strdup (own definition and from strings.h), shm_strdup, dupl_string, and in some places of course several macros with the same purpose. As i mentioned before, the same problem exist with some of our internal modules. It is the same with the str_fixup functions (char* to str) and in the db layer with the converter functions, i allready submit a patch some time ago for this. The current situation is in my opinion not optimal from the maintenace point of view, it also increases the code size and could decrease the performance. I'm willing to work on this, i only need to know if this work is welcome. One normal str_dup function that uses string.h, and one for shm_memory should be sufficient. Furthermore i propose a change to the contribution rules on the OpenSER webpage, that any new modules should use the existing utility functions in the core. Part of the problem is the lack of a good hacking guide for OpenSER, many module authors don't know probably about the e.g. ut.h header file. Cheers, Henning _______________________________________________ Devel mailing list Devel@openser.org http://openser.org/cgi-bin/mailman/listinfo/devel
