Sarathy wrote: !That's not right according to the documentation in config_h.SH and !Porting/Glossary. I've fixed Config.t instead, since the same issue !shows up on windows. If the test passed on UNIXes, it sounds like !Configure is not doing the right thing here?
In the [EMAIL PROTECTED] kit Craig's patch has been applied and the new version of the lib/Config.t test also passes on VMS. Here is a possible doc patch for config_h.SH. If it is decided that this is acceptable (in light of the controversy I realize that is not assured) then porting/Glossary ought to be re-generated: --- config_h.SH;-1 Mon Mar 17 09:36:27 2003 +++ config_h.SH Mon Mar 17 13:49:24 2003 @@ -2212,8 +2212,6 @@ * Gaps in the signal numbers (up to NSIG) are filled in with NUMnn, * etc., where nn is the actual signal number (e.g. NUM37). * The signal number for sig_name[i] is stored in sig_num[i]. - * The last element is 0 to terminate the list with a NULL. This - * corresponds to the 0 at the end of the sig_name_init list. * Note that this variable is initialized from the sig_name_init, * not from sig_name (which is unused). */ @@ -2228,14 +2226,12 @@ * Duplicates are allowed, but are moved to the end of the list. * The signal number corresponding to sig_name[i] is sig_number[i]. * if (i < NSIG) then sig_number[i] == i. - * The last element is 0, corresponding to the 0 at the end of - * the sig_name_init list. * Note that this variable is initialized from the sig_num_init, * not from sig_num (which is unused). */ /* SIG_SIZE: * This variable contains the number of elements of the sig_name - * and sig_num arrays, excluding the final NULL entry. + * and sig_num arrays. */ #define SIG_NAME $sig_name_init /**/ #define SIG_NUM $sig_num_init /**/ End of Patch. Peter Prymmer
