On Thu, Sep 20, 2007 at 06:51:45AM +0200, Sam Ravnborg wrote: [] > > > There are several small things that are not addressed in todays > > > implementation and that will be fixed one way or the other. > > > > Did they get addressed year later or not? What they were (are)? > > Nope. > I have not visited the -combine thing since, neither has anyone alse.
As far as i can see, gcc still in trouble. I'm not sure what version/branch will have all problems fixed. What i think, is making simple text compilation (with some C language altering, like externs, etc.)of build tokens, based on include and build dependencies, and then passing one file to gcc. C Language and text processing tools were designed decades ago, but i don't see harmonic cooperation against modern demands. I don't know what performance penalties this may cause, but still trying isn't that bad. This depends on `sed` tool mainly, but maybe my optimized sed scripting and GNU sed's performance will gain against multiple running of gcc, which parses all its arguments and includes every single run, and extensively uses (very cheap in Linux) forking/dynamic loading. > If you plan to do so then please do it in small > incremental steps each introducing *minimal* changes. I'm trying to design something different (maybe not so new at all), as you probably remember. I read all that (flame) threads back in 1999, 2001, 2002 about build system. Trying to get all experience from most kernel- build related stories. Now it's dead topic as it seems, but anyway. _____ ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ kbuild-devel mailing list kbuild-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/kbuild-devel