> > Hi. > I have (a long time ago) created an AAA module based originally on > Apache2::AuthCookie (a copy and rewrite, not a sub-class). Now I need to > adapt this to Apache 2.4. > I have read all the docs at > https://metacpan.org/source/MSCHOUT/Apache-AuthCookie-3.24/lib/Apache2_4/AuthCookie.pm, > and followed the related correspondence on this list. > My problem is that I would like to keep a single perl source, for both > Apache 2.2 and Apache 2.4. > As far as I can tell so far, this should not be a big problem, except for > the following difference : > > In the Apache 2.2 version, I have this statement : > > use Apache2::Const -compile => qw(:common :http :methods :override :proxy); > > and in the Apache 2.4 version, I must have this statement or it will not > compile : > > use Apache2::Const -compile => qw(:common :http :methods :override :proxy > AUTHZ_GRANTED AUTHZ_DENIED AUTHZ_DENIED_NO_USER); > > I'm not very familiar with require vs use, but could you use a begin block, Apache2::ServerUtil::get_server_version() and then if/else based on the version number?
Lathan Bidwell > (That is because these new AUTHZ_* constants must be used in the new > "AuthzProvider" subs). > > I could of course hack the values of these constants under 2.4, and use > their values directly in my module, but that is rather "inelegant". > > My question is thus : would there be a way to make the import of these > constants conditional on the Apache version the module is running under ? > (In other words e.g. define them with value 0 initially (for Apache 2.2, > since they are never used there), and override them if under Apache 2.4 ? > > Or, to put this another way : imagining that there are already some > servers, currently running Apache 2.2 with, say, mod_perl 2.0.7, and on > which the current (2.2-only) version of my Auth module is running. > If I ever update these servers with the latest (2.4-compatible) version of > my module, including the above extra imported constants, that updated > module will not compile anymore on those systems (because these constants > are not defined there). > Is there a way to avoid this incompatibility ? > > I tried to phrase the above as clearly as possible, but if it is still > confusing, I can try again.. > > Thanks > >
