On 5/19/05, John Torjo <john.lists@torjo.com> wrote:
In order to make the Logging lib as usable as it can be, I'd like to know what is *your* scenario, when using a Logging lib.
Not sure if this qualifies as a "scenario", but these are my high-level requirements for a logging framework: * Easy to select/change output destinations (e.g. stdout/stderr, file, syslog, etc). Output can be sent to multiple destinations if desired. Default will send output to stdout if nothing is explicitly specified by the user. * Configurable output format. User should be able to associate a format with an output destination (where applicable - it may not make sense in all cases). The current modifier chain does this reasonably well. * Channels or topics or keywords. I think various people have asked for this by one or more of these names, and I like the idea as well. This can be implemented with separate loggers in the current implementation, but I'm not sure how I feel about that. It doesn't feel right to me. * Severity Levels. Topics have a configurable threshold at or above which they will output a message; messages of a lower severity are ignored. Users specify the severity level with each message. This is orthogonal to the channel/topic/keyword, but should be configurable by c/t/k or with wildcards (e.g. logger.threshold ("*", logger::levels::warning); logger.threshold ("net", logger::levels::debug)). What about user-defined levels? * Thread-safe. Two threads should never have their messages intermixed as they would with raw ostream<< operations. * Logfile rotation. For file-based logs, there should be a way to configure the framework to . The appenders you included in 1.3.3 implement this. * Simple interface using ostream::operator<<. If my classes have streaming support, they should be log-able. Something like: BOOST_LOG (logger, topic, level) << message << goes << here or BOOST_LOG (logger, topic, level, message << goes << here) * Structured log records. In order to implement thread-safety, there must understandably be some level of buffering. I like the current ts_appender with its timed batching of writes, but you lose some information like the name of the logger to which the message was originally sent. I think it is imperative that a logging "call" is able to capture a precise timestamp, thread ID, topic, severity, and __FILE__/__LINE__ info at time the message is created. If a data structure is used to represent the message instead of just a std::string, this information can be captured. Some folks might not like the overhead associated with timestamping etc., so perhaps this could be provided through an alternate interface, e.g.: BOOST_LOG_RECORD (topic, level, message << goes << here) would create a structured log_record object fully populated with thread ID, timestamp, topic, level, etc but BOOST_LOG (topic, level, ...) would create a log_record with only the message, __FILE__ and __LINE__ data initialized (the latter two should cost almost nothing). * Setup. I'm not enamored of the requirement of declaring & defining loggers (or using scoped loggers). If the channel & severity concepts become first class citizens, is there any reason not to have a singleton logger? Or perhaps a singleton logger could be made available, and users could provide their own loggers instead if they desired. -- Caleb Epstein caleb dot epstein at gmail dot com