Stefan, The script could be repurposed to create a bloat-o-meter, but that wouldn't reveal the cost of the individual patches, which I'm interested in.
Thomas, I had hoped that the script could do incremental rebuilds fast without ccache using "make ; git checkout <commit_id> ; make ; ..." but it seems something is conservatively forcing full rebuilds when I do "git checkout". So, ccache could help. However, the script does cache full build results already, so it already does better than ccache in some cases by entirely avoiding rebuilds. Piotr, with appropriate effort, leveraging the script result caching, the 3300 commits need not be an issue, except for the first generated plot, which would take >10hrs. By the way, is there a reason gpxe doesn't use much faster conservative dependency generation? For example: > > time find . -name '*.[chS]' | make_depend > bar > > real 0m0.044s > user 0m0.020s > sys 0m0.020s --Glenn _______________________________________________ gPXE-devel mailing list [email protected] http://etherboot.org/mailman/listinfo/gpxe-devel
