Gavin Maltby wrote: > On 06/09/06 11:31, Darren Reed wrote: > >> To bring the syslogd shipped with Solaris up to a level that is >> more in line with what is found in other systems today, I'd like >> to propose a project to upgrade it. >> >> Tasks currently scoped out for this project include: >> - introduce some of the more modern aspects of syslog.conf such as >> the use of '*' and '!' >> - filling out the list of facilities defined where there is common >> usage of undefined ones in the opensource community >> - implement the IETF TLS transport mapping for syslog >> - supporting explicit IPv6 destinations >> - adding support for specifying extra log devices > > > I'd like to see the syslog files be structured (binary) files.
If there is any evolution of the log file format, it will be to use XML. XML and structured data have featured heavily in discussions about advancing syslog and the benefits are clear: data gets typed. If the protocol data is formatted this way then there is little sense in writing it out in a binary format. If/when that happens it may then be appropriate to look at the syslog(3) interface to see if changes could be made there to make it easier to get data into syslog and out as XML. Moinak Ghosh wrote: > Darren Reed wrote: > >> To bring the syslogd shipped with Solaris up to a level that is >> more in line with what is found in other systems today, I'd like >> to propose a project to upgrade it. >> >> Tasks currently scoped out for this project include: >> - introduce some of the more modern aspects of syslog.conf such as >> the use of '*' and '!' >> - filling out the list of facilities defined where there is common >> usage of undefined ones in the opensource community >> - implement the IETF TLS transport mapping for syslog >> - supporting explicit IPv6 destinations >> - adding support for specifying extra log devices > > > +1 > > Also I'd like suggest augmenting the API with an improved variant > of openlog() (openlog_r() ?) that fixes current issues relating to thread > safety and dlopen risks. I'll add making the syslog libc API thread-safe as this is a no brainer for Solaris. I haven't looked into this problem in any depth so I can't comment (and don't want to commit to anything) in detail about what is or isn't required. Darren
