NightStrike wrote: > This may be a totally wrong question, but I'll ask anyway :) > > I see that Makefile.in is generated using autogen from Makefile.def > and Makefile.tpl. Why does gcc use this method instead of automake > with Makefile.am? The toplevel Makefile doesn't actually build any code, it's just used to control the bootstrap process and is primarily responsible for descending into a number of subdirectories in a certain and careful order while passing down large number of options. So most of what Automake brings (e.g. automatic dependency generation as a side effect of compilation) would be of no use. On the other hand, if you look at the generated Makefile.in there is a vast amount of duplication due to the fact that these toplevel files are shared across a large number of projects (gcc, gdb, newlib, binutils, cygwin, mingw, ...) and because the bootstrap process consists of a number of stages. When you take the cartesian product of these dozens of subdirectories to descend into crossed with up to four stages combined with several targets for each subdir (configure, clean, check, html, dvi, info, ...) you get an enormous number of targets. This kind of rote duplication is better suited to a pure text processing language like autogen. Brian
