Qpid Proton C API  0.33.0
logger.h File Reference

Facility for logging messages. More...

#include <proton/import_export.h>
#include <proton/object.h>
#include <stdarg.h>
#include <stdint.h>

Go to the source code of this file.

Typedefs

typedef struct pn_logger_t pn_logger_t
 The logger object allows library logging to be controlled.
 
typedef void(* pn_log_sink_t) (intptr_t sink_context, pn_log_subsystem_t subsystem, pn_log_level_t severity, const char *message)
 Callback for sinking logger messages.
 

Enumerations

enum  pn_log_subsystem_t {
  PN_SUBSYSTEM_NONE, PN_SUBSYSTEM_MEMORY, PN_SUBSYSTEM_IO, PN_SUBSYSTEM_EVENT,
  PN_SUBSYSTEM_AMQP, PN_SUBSYSTEM_SSL, PN_SUBSYSTEM_SASL, PN_SUBSYSTEM_BINDING,
  PN_SUBSYSTEM_ALL
}
 Definitions for different subsystems that can log messages. More...
 
enum  pn_log_level_t {
  PN_LEVEL_NONE, PN_LEVEL_CRITICAL, PN_LEVEL_ERROR, PN_LEVEL_WARNING,
  PN_LEVEL_INFO, PN_LEVEL_DEBUG, PN_LEVEL_TRACE, PN_LEVEL_FRAME,
  PN_LEVEL_RAW, PN_LEVEL_ALL
}
 Definitions for different severities of log messages Note that these are exclusive bits so that you can specify multiple severities to filter. More...
 

Functions

pn_logger_tpn_default_logger (void)
 Return the default library logger. More...
 
const char * pn_logger_level_name (pn_log_level_t level)
 Get a human readable name for a logger severity. More...
 
const char * pn_logger_subsystem_name (pn_log_subsystem_t subsystem)
 Get a human readable name for a logger subsystem. More...
 
void pn_logger_set_mask (pn_logger_t *logger, uint16_t subsystem, uint16_t level)
 Set a logger's tracing flags. More...
 
void pn_logger_reset_mask (pn_logger_t *logger, uint16_t subsystem, uint16_t level)
 Clear a logger's tracing flags. More...
 
void pn_logger_set_log_sink (pn_logger_t *logger, pn_log_sink_t sink, intptr_t sink_context)
 Set the tracing function used by a logger. More...
 
pn_log_sink_t pn_logger_get_log_sink (pn_logger_t *logger)
 Get the tracing function used by a logger. More...
 
intptr_t pn_logger_get_log_sink_context (pn_logger_t *logger)
 Get the sink context used by a logger. More...
 
void pn_logger_logf (pn_logger_t *logger, pn_log_subsystem_t subsystem, pn_log_level_t level, const char *fmt,...)
 Log a printf formatted message using the logger. More...
 

Detailed Description

Facility for logging messages.

The proton library has internal logging which provides information about the functioning of the library. This Logger system (see pn_logger_t) allows applications to customize how the logging is recorded and output.

Each logged message has an attached level (see pn_log_level_t) which indicates how important the message is; and also has an attached subsystem (see pn_log_subsystem_t) which indicates which part of proton is producing# the log message. The levels go from "Critical" which indicates a condition the library cannot recover from.

Applications receive log messages by registering a callback function (pn_log_sink_t). This receives the logged message and other information and allows the application to consume the logged messages as it wants. Applications can filter the messages that they receive by setting bit masks in the logger. They can set the logger filtering both by subsystem and by level. Additionally, since the callback contains both the subsystem and level information for each logged message, applications can impose further, more complex, filters of their own.

Each application will have a default logger which can be retrieved with pn_default_logger. The default logger is used as the template for every other logger that is created. So an efficient way to configure logging is to configure the default logger at the very start of the application before any other loggers get created.

Loggers are associated with different proton objects, primarily the transport object (pn_transport_t) and each logger controls the logging for the associated object. This means that for example in order to control the logging for an AMQP connection you need to acquire the logger object from the transport object using pn_transport_logger and to configure that.

Initially the defaults are to log every subsystem but not any level (except 'Critical' which is always logged). This means in effect that it is only necessary to turn on the log levels that are interesting and all subsystems will log. Of course you can turn off subsystems that are not interesting or are too verbose.

There is also a default log sink if the application does not register their own, but logging is turned on - this will output the log message to standard error.