RE: [logs] Syslog payload format

Previous message: Frank O'Dwyer: "RE: [logs] Syslog payload format"
In reply to: Marcus J. Ranum: "Re: [logs] Syslog payload format"
Next in thread: wolfgangat_private: "Re: [logs] Syslog payload format"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

fodat_private

Marcus J. Ranum wrote:
> Darren Reed wrote:
[...]
> >Now that I think about it, I think a requirement must be that it
> >provides a syslog(3) interface for old applications.
>
> You're probably right. Of course if one provided
> such an interface there'd be no motivation to ever
> change it, and we'd be (basically) stuck with syslog
> forever.

You could easily arrange for the compiler to only provide the old interface
if -DNEW_SYSLOG wasn't on, and also spit out warning messages ("syslog(3) is
broken. please use -DNEW_SYSLOG, see such and such for recommended usage")
in that case.

That way old programs continue to compile and work without change, but you
still give a pretty clear message that the old interface is deprecated.

After a while (i.e. if the new API caught on) you could then reverse the
behaviour so that you had to explicitly do something like -DBROKEN_SYSLOG to
get the old interface.

Cheers,
Frank

_______________________________________________
LogAnalysis mailing list
LogAnalysisat_private
http://lists.shmoo.com/mailman/listinfo/loganalysis