On Sun, Nov 13, 2011 at 07:39:33PM +0100, David Kastrup wrote: > Graham Percival <gra...@percival-music.ca> writes: > > > It's a shame that we haven't implemented the logging system > > accepted by GOP... 9, was it? If we had that, it would take about > > 30 seconds to figure out **exactly** which piece of lilypond input > > was dying. > > Sounds to me like parallel build is responsible again.
But a good logging system would still tell us what was being built. If two threads were trying to build the same file, then we know we need to add a lockfile or something like that. If only one thread was building that file and it still dies, then we know it's just that input. Now that I think aboud it, the parallel build thing seems very plausible to me. Multiple languages can be building exactly the same snippet at the same time, and I could well imagine that this will cause boom. ... but without a logging system, we might as well be theorizing about the universe before the big bang. Potentially fun, but not likely to be useful for any practical purpose. - Graham _______________________________________________ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
