On Fri, Sep 22, 2017 at 2:31 PM, Joe Orton <[email protected]> wrote: > On Fri, Sep 22, 2017 at 11:58:53AM -0000, [email protected] wrote: >> --- httpd/httpd/trunk/include/ap_mmn.h (original) >> +++ httpd/httpd/trunk/include/ap_mmn.h Fri Sep 22 11:58:53 2017 > ... >> @@ -562,7 +563,7 @@ >> #ifndef MODULE_MAGIC_NUMBER_MAJOR >> #define MODULE_MAGIC_NUMBER_MAJOR 20161018 >> #endif >> -#define MODULE_MAGIC_NUMBER_MINOR 6 /* 0...n */ >> +#define MODULE_MAGIC_NUMBER_MINOR 7 /* 0...n */ > > Since this adds fields to module_struct, not sure how it's anything but > a major bump (not backwards-compatible)?
We started to discuss it in the "SSLSrvConfigRec shared" thread, but it's indeed more appriopriate here I think. Usually a new field *at the end*" of structs is a minor bump only... For module_struct, I don't think it's that sensible because modules are usually declared globally, hence existing modules recompiled against latest core will have their 'flags' field initialized to zero. I also took care (hopefully, r1809305) in the core to not try to access the 'flags' field if the module was not compiled against latest core. So it should be safe, unless e.g. one managed to create an array of module_structs... But ISTM that we're allowed to minor bump only here. Regards, Yann.
