[87295] trunk/dports/databases

[Ryan Schmidt]

On Nov 16, 2011, at 12:23, jwa at macports.org wrote:

> Revision: 87295
>          http://trac.macports.org/changeset/87295
> Author:   jwa at macports.org
> Date:     2011-11-16 10:23:45 -0800 (Wed, 16 Nov 2011)
> Log Message:
> -----------
> new port, at 5.0.1

Remember to begin your commit message with the port name.


> Added Paths:
> -----------
>    trunk/dports/databases/libgda5/
>    trunk/dports/databases/libgda5/Portfile


> +if (![variant_isset without_bdb]) {
> +    if !([file exists ${prefix}/include/db46/db.h]) {
> +        depends_lib-append port:db47
> +    } elseif ([file exists ${prefix}/include/db47/db.h])  {
> +        depends_lib-append port:db47
> +    } else {
> +        depends_lib-append port:db46
> +    }
> +}

I'm surprised this works syntactically; I've never seen parentheses used in Tcl in "if" statements. Everywhere else, we use curly braces.

I'm also not sure it's a good idea to vary the port's behavior based on what's already installed on the user's system; that's something we usually avoid. It's not like BerkeleyDB is a huge dependency that would take a long time to build or take a lot of disk space. Why not just unconditionally depend on the latest version that will work? The latest we have in MacPorts is db52.


> +variant with_mysql5 \
> +    description {support for current MySQL 5.x} {
> +	depends_lib-append	path:bin/mysql_config5:mysql5
> +	configure.env-append	\
> +		MYSQL_CONFIG=${prefix}/lib/mysql5/bin/mysql_config
> +	configure.args-delete   --without-mysql
> +	configure.args-append	--with-mysql=${prefix}/lib/mysql5
> +	configure.cppflags-append	"-I${prefix}/include/mysql5/mysql"
> +}
> +
> +variant with_postgresql83 conflicts with_postgresql84 with_postgresql90 \
> +    description {support for PostgreSQL 8.3.x} {
> +	configure.cppflags-append	"-I${prefix}/include/postgresql83"
> +	depends_lib-append	port:postgresql83
> +	configure.args-delete   --without-postgres
> +	configure.args-append	--with-postgres=${prefix}/lib/postgresql83
> +}
> +
> +variant with_postgresql84 conflicts with_postgresql83 with_postgresql90 \
> +    description {support for PostgreSQL 8.4.x} {
> +	configure.cppflags-append	"-I${prefix}/include/postgresql84"
> +	depends_lib-append	port:postgresql84
> +	configure.args-delete   --without-postgres
> +	configure.args-append	--with-postgres=${prefix}/lib/postgresql84
> +}
> +
> +variant with_postgresql90 conflicts with_postgresql83 with_postgresql84 \
> +    description {support for PostgreSQL 9.0.x} {
> +	configure.cppflags-append	"-I${prefix}/include/postgresql90"
> +	depends_lib-append	port:postgresql90
> +	configure.args-delete   --without-postgres
> +	configure.args-append	--with-postgres=${prefix}/lib/postgresql90
> +}
> +
> +variant without_bdb \
> +    description {remove support for Berkeley DB} {
> +	configure.args-delete   --with-bdb=${prefix}
> +	configure.args-append	--without-bdb
> +}

We no longer want variant names starting with "with_", "without_", "no_", etc. These variants should be called "mysql5", "postgresql83", "postgresql84", "postgresql90" and "bdb", respectively. The bdb variant's functionality would be the reverse of what you currently have in the without_bdb variant. And you'd then make bdb a default variant with:

default_variants +bdb

And to keep everything related to bdb together, the "if" statement stuff from above that sets the bdb dependency should be inside the new bdb variant.


> +post-activate {
> +    system "${prefix}/bin/gtk-update-icon-cache -f -t ${prefix}/share/icons/hicolor"
> +}

If it's important to update the icon cache after activation, is it also important to do so after deactivation? (I'm not sure, just asking.)