On 11/05/2010 13:29, Björn Michaelsen wrote: > This leads to some "interesting" constructions, for example on OSX > where the linker needs to know about the runtime locations of libs it > links against. Assuming this could simply be abstracted away with > built-in CMake facilities is IMHO quite naive. And there are certainly > a lot of other pitfalls to be found, so that the assumption that it > would be easier to use CMake just isnt valid.
hmm... ideally the solver would have the same directory layout as the installed office, in which case linking on OSX would probably be a lot easier :) but of course there is the point again that you shouldn't change everything at once, because you'll never finish... with the GBuild stuff it's possible to put in the same workaround that is used in the old build system (horribly ugly perl hackery), and then later, after everything is migrated to GBuild, we can have an additional step where we adapt the layout of the solver and get rid of the ugly hacks. this seems like a practical, incremental migration path. > In addition, for stuff like our own localization toolchain etc. we cant > possibly hope for any support from CMake, so it makes little difference. > > Best Regards, > > Bjoern -- "Odious ideas are not entitled to hide from criticism behind the human shield of their believers' feelings." -- RMS --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
