labath wrote: > Regarding limiting the output, I'm happy to draw the line at > `LLDB_LOG_ERRORV`. The system log is pretty cheap on our platform and I > assumed it's the same on Linux, so personally I would hope it doesn't deter > anyone from using `LLDB_LOG_ERROR`.
>From the looks of things, these "system logs" will end up somewhere in >`/var/log` (depending on theexact system logger configuration and stuff) on >linux, but I associate these logs mostly with daemon processes which don't >really have a way to notify the user (because there isn't one) of any problems >they encounter. Lldb has ample opportunity to do that, so using this as a sink >for messages we haven't bothered to propagate to the user doesn't seem >completely appropriate to me (and its not what I imagined when I saw the first >patch in this series). For example, on linux, these system logs are normally only readable by root, so a normal user wouldn't even be able to see what is being logged, and the log messages would get interspersed with messages from daemons or programs like su, cron, NetworkManager, etc. (it's possible to configure this differently, but this also is something only system administrators can do). I actually don't remember seeing any non-daemon process write to these logs. I suppose the situation could be different on darwin, and maybe this just means that the linux/posix implementation of the "system log" is wrong (in this brave new world, at least), but I also don't know what would be a good implementation of this function. @DavidSpickett, do you have any thoughts on this? :) https://github.com/llvm/llvm-project/pull/111911 _______________________________________________ lldb-commits mailing list lldb-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
