2006/6/14, Mark Hindess <[EMAIL PROTECTED]>:
On 14 June 2006 at 7:24, Geir Magnusson Jr <[EMAIL PROTECTED]> wrote: > > Oliver Deakin wrote: > > I have a couple of questions about location of makefiles and makefile > > includes > > > > 1) Currently the makefiles for the modules are stored underneath the > > platform > > and component they relate to. For example, the luni makefiles are situated > > at native-src/linux.IA32/luni/makefile and > > native-src/win.IA32/luni/makefile. > > > > Once I move the luni native code into the luni module, its code > > layout will look like: > > > > modules/luni/src/main/native/luni/ > > |---shared > > |---linux > > \---windows > > > > But where should the platform specific makefiles go? > > > > I think we have two choices here - put the linux one in the linux dir, > > and similar for the windows version (as it is now), or put them both > > under the modules/luni/src/main/native/luni/ directory and rename them > > something like makefile.linux and makefile.windows. > > > > Personally Im happy to go with the first option and keep each makefile > > under its corresponding platform directory until we have reason > > to change it. Just thought Id gauge if anyone has any preference. > > I agree. I'm also wondering how painful it would be to switch to > something other than NMAKE as it seems pretty braindead. I agree. Besides it's quite easy to change later - if we find something better than nmake for instance. (I loathe nmake because it doesn't even have some of the most trivial features for variable manipulation.) I did try using the mingw toolset but didn't get very far but I'll try looking at it when we are modularised - one module at a time.
Also if we decide to go with gmake then later we may add a configure script to find dependencies like apr, icu, icu4j, log4cxx, zlib and others provided by the system installation instead of downloading them. -- Gregory Shimansky, Intel Middleware Products Division
