On Wed, Dec 5, 2012 at 2:12 PM, Jonathan "Duke" Leto <jonat...@leto.net> wrote: > This article by the main developer behind Redis has some very interesting > ideas about practical ways of improving software quality: > > http://antirez.com/news/43 > > I especially like that he has added various amounts of detailed output which > often allows him to debug a situation without asking a user for a core > dump/etc. > > Parrot could definitely benefit from implementing something similar.
Nice. We've already got a backtrace on most platforms, which is better than most others have. But not in the weird platforms we support, and with which we have problems with sometimes. Like the recent /proc/cpuinfo exit on darwin without procfs. But this was a normal exit(1) so only strace/dtruss helps. But there are nice other ideas, like printing more environmental stuff, client info (argv), env values, register values,/proc/pid/maps we could add to our backtrace or some kind of parrot crash log. ramtest, memcheck added to --gc-debug --leak-test? hmm. why not. but it might cause more memory footprint for a normal vm most users will most likely not want to carry around. I love the link in the comments about this clang plugin: https://github.com/yotamr/traces to generate better API traces. This also traces deep, not only the flat op calls. I find it more disturbing that we still not generate manpages. Which cmdline option? $ man parrot oops -- Reini Urban http://cpanel.net/ http://www.perl-compiler.org/ _______________________________________________ http://lists.parrot.org/mailman/listinfo/parrot-dev
