Patches for OpenSolaris

[Dan Meltzer]

On Wed, Jan 21, 2009 at 2:33 PM, Adriaan de Groot <groot at kde.org> wrote:
> Here at CampKDE I showed the Amarok dudes how it would run on OpenSolaris; it
> does take a fistful of patches (mostly simple ones). Casey suggested I
> actually send them in instead of blogging only :)
>
> The three const-match patches make declaration and definition of various
> methods match in terms of constness; gcc thinks this is fine, but for
> compliers that consider constness in resolution or linking, this is a problem.
>
> Free is in <stdlib.h> and doesn't necessarily live in the :: namespace.
>
> The findmysql one is possibly not suitable for inclusion -- it adds some more
> search paths and things to find MySQL in the default install locations on
> Solaris.

Our findmysql.cmake is pretty nutty.  I think we need to list standard
locations and expect places that don't install it to the standard
location to simply do cmake -DMYSQL_LIBRARIES=blahblahblah.  This
seems to be the standard route to handle this, I'm not sure why
findmysql.cmake is being treated differently (not by you, by us...)
>
> The md5 code uses the linux kernel types u_int32_t instead of standard integer
> types like uint32_t.

This is external code... though I'm not sure if it's actually being
maintained, if it is you might want to see about getting it included
there (though the last time we updated this code was a forever ago...
>
> namespace foo { extern bar; } gives me a foo::bar symbol instead of a ::bar
> symbol, so the namespace patch moves one extern around.
>
> The scriptengine has some constness issues and a ; too many. I think Casey has
> already committed that last one.

These should be submitted to the qtscriptgenerator project upstream as
well, or they will go away the next time we update the
scriptgenerator.
>
> The amarok-wide patch is rather klumky, since it just forces in the Solaris
> wide character functions instead of doing some suitable conditional
> compilation. I'd be ok with maintaing this one separately.
>
> The DEBUG_BLOCK macro uses the gcc-ism __PRETTY_FUNCTION__; see qdebug for one
> way of dealing with this instead. I've gone for the boring approach of
> #defining it to a string that is semi-sort-of-meaningful.
>
> Finally, no-webkit removes WebKit from linking, because it conflicts with kjs
> in KDE -- those are harmless collissions, i think, but it's annoying to see
> all those show up on linking.
>
> [ade]
>
> _______________________________________________
> Amarok-devel mailing list
> Amarok-devel at kde.org
> https://mail.kde.org/mailman/listinfo/amarok-devel
>
>