On 15.08.08 00:13:49, Alexander Neundorf wrote: > On Thursday 07 August 2008, Andreas Pakulat wrote: > > On 07.08.08 22:11:11, Thiago Macieira wrote: > > > Andreas Pakulat wrote: > > > >This has a problem, I can't depend on KDE 4.2.0 until its released and > > > >hence modules in trunk/ cannot have a dependency on kdelibs from trunk/ > > > > > > > >For example kdevplatform now has a hard dependency on kdelibs from > > > >trunk/, so what I need is require KDE >= 4.1.60 (current version used in > > > >trunk/) not 4.2.0 as that doesn't exist yet and 4.1.0 doesn't work > > > >anymore. > > > > > > Right, sorry. I hadn't thought of pre-releases. > > > > Thanks for doing the logic-thinking for me :) Patch attached, it uses a > > negation of your pseudo-code to avoid having to create new variables. > > > > Andreas > > Looks good from the technical side. > Still there is one thing: > right now the way to specify the required KDE version is to set the > KDE_MIN_VERSION variable before calling find_package(KDE4). > > This patch now adds a second way how to specify the required version. > Pro: > -this is how find_package() now supports specifying the minimum version > number, i.e. in some way the officially recommended way
Another pro is that we don't have to take care of moving a whole block (variable-setting+find_package) when reorganizing a module's CMakeLists.txt. > Con: > -with the patch there are now two ways how to do the same thing (and the old > way has to stay, in order to stay source compatible) Well, while I agree its better to leave it in, the variable isn't documented and hence is IMHO an internal variable - bound to change at any time. Just like in code anything that has no api dox is bound to change at any time. > -several other modules also support FOO_MIN_VERSION or similar variables, so > using such a variable is currently kind of de-facto standard for specifying > the minimum version. Well, I think those modules support a version variable only because there was no other way until now. If we'd always just stay with what we have we'd never make any progress at all :) So to conclude: I don't think the two con's weigh heavy enough not to apply the patch. Andreas -- You will be awarded some great honor. _______________________________________________ Kde-buildsystem mailing list [email protected] https://mail.kde.org/mailman/listinfo/kde-buildsystem
