There may have been code that compiled using those headers, but it
wouldn't have linked.
Found a bug perusing SyslogAppender (notice the repeated use of LOCAL1):
else if (StringHelper::equalsIgnoreCase(s,
LOG4CXX_STR("LOCAL0"), LOG4CXX_STR("local0")))
{
return LOG_LOCAL0;
}
else if (StringHelper::equalsIgnoreCase(s,
LOG4CXX_STR("LOCAL1"), LOG4CXX_STR("local1")))
{
return LOG_LOCAL1;
}
else if (StringHelper::equalsIgnoreCase(s,
LOG4CXX_STR("LOCAL1"), LOG4CXX_STR("local2")))
{
return LOG_LOCAL2;
}
else if (StringHelper::equalsIgnoreCase(s,
LOG4CXX_STR("LOCAL1"), LOG4CXX_STR("local3")))
{
return LOG_LOCAL3;
}
else if (StringHelper::equalsIgnoreCase(s,
LOG4CXX_STR("LOCAL1"), LOG4CXX_STR("local4")))
{
return LOG_LOCAL4;
}
else if (StringHelper::equalsIgnoreCase(s,
LOG4CXX_STR("LOCAL1"), LOG4CXX_STR("local5")))
{
return LOG_LOCAL5;
}
else if (StringHelper::equalsIgnoreCase(s,
LOG4CXX_STR("LOCAL1"), LOG4CXX_STR("local6")))
{
return LOG_LOCAL6;
}
else if (StringHelper::equalsIgnoreCase(s,
LOG4CXX_STR("LOCAL1"), LOG4CXX_STR("local7")))
{
return LOG_LOCAL7;
}
On Thu, May 1, 2008 at 11:22 AM, Curt Arnold <[EMAIL PROTECTED]> wrote:
>
>
> On May 1, 2008, at 9:45 AM, Dale King wrote:
>
>
> > It looks like there are a few filters that have header files, but no
> > source files. In particular, AndFilter, ExpressionFilter,
> > LocationInfoFilter, MapFilter and PropertyFilter appear to not be
> > implemented.
> >
> > If there is no implementation I don't think the headers should be
> > included as this will confuse someone that sees them in the
> > documentation and tries to use them.
> >
> > -- Dale King
> >
>
>
> I've added LOGCXX-276 for this issue.
>
> The definitions were apparently added while porting over the log4j 1.3
> rolling file appender, but the implementations were not exercised by the
> unit tests and the lack of implementation was not noticed.
>
> Removing the headers could cause a program that successfully compiled to
> fail to compile. Would have been okay to pull them before the 0.10.0
> release, but now we need to keep compatibility until we are ready to jump to
> 0.11.x. The best way to fix the problem is to add the implementation in or
> if that is not possible to note in the headers that they are unimplemented.
>
--
Dale King
