On 2/14/11 3:22 PM, retard wrote:
Your obsession with fast compile times is incomprehensible. It doesn't have any relevance in the projects I'm talking about. On multicore 'make - jN', distcc& low cost clusters, and incremental compilation already mitigate most of the issues. LLVM is also supposed to compile large projects faster than the 'legacy' gcc. There are also faster linkers than GNU ld. If you're really obsessed with compile times, there are far better languages such as D. The extensive optimizations and fast compile times have an inverse correlation. Of course your compiler compiles faster if it optimizes less. What's the point here? All your examples and stories are from 1980's and 1990's. Any idea how well dmc fares against latest Intel / Microsoft / GNU compilers?
I work on a >1M LOC C++ project and using distcc with 4 nodes and ccache. Unfortunately, it is not enough. Yes, there are various cases where runtime performance matters a lot. But compile time performance of C++ is a huge problem. I am glad that Walter cares about this.
The point about optimizations vs compile time seems to be a valid one. However, even without optimizations turned on gcc sucks big time w.r.t. compilation time. And most of the time is being spent in parsing gazillion number of headers. I did not have a chance to work with Intel's and MS's compilers.
