Re: [PATCH] wayland: introduce logging with syslog

2011-07-25 Thread Tiago Vignatti
On 07/25/2011 06:18 PM, Kristian Høgsberg wrote: On Fri, Jul 22, 2011 at 1:01 PM, Tiago Vignatti wrote: Three level of messages are possible, which are sent to syslog: WL_INFO, WL_WARNING and WL_ERROR; the last two are printed also to stderr due sanity. The only interface for the compositor an

Re: [PATCH] wayland: introduce logging with syslog

2011-07-25 Thread Kristian Høgsberg
On Fri, Jul 22, 2011 at 1:01 PM, Tiago Vignatti wrote: > Three level of messages are possible, which are sent to syslog: WL_INFO, > WL_WARNING and WL_ERROR; the last two are printed also to stderr due > sanity. The only interface for the compositor and clients is > >    void wl_log(int level, cons

Re: [RFC] implement log facility using syslog

2011-07-25 Thread Tiago Vignatti
On 07/23/2011 01:32 AM, Christian Hergert wrote: Can syslog() potentially block? If so, that might be an issue of concern. I don't think it is, at least officially stated, that syslog() is re-entrant safe (which may be of concern with regard to how signals are handled). well, this depends the

Re: [RFC] implement log facility using syslog

2011-07-25 Thread Tiago Vignatti
On 07/23/2011 12:34 AM, Marty Jack wrote: The primary reason you wouldn't use syslog to log user messages is that it opens the possibility for an unprivileged user process to maliciously consume the root partition. Logging to the equivalent of ~/.Xsession-errors is the right way to go becau