You are viewing a plain text version of this content. The canonical link for it is here.
Posted to log4cxx-dev@logging.apache.org by Tommi Mäkitalo <to...@epgmbh.de> on 2004/12/28 23:07:31 UTC

missing header?

Hi Curt,

I just checked out a fresh version and realized, that you checked in a 
nonworking src/transcoder.cpp. You added a "#iinclude <log4cxx/config.h>". 
Did you change config_auto.h to config.h?


Tommi

Re: missing header?

Posted by Tommi Mäkitalo <to...@epgmbh.de>.
>
> The build had been tweaked so the output of configure is processed to
> change "HAVE_WHATEVER" into "LOG4CXX_HAVE_WHATEVER".  That processing
> is no longer needed since we do not include any of the "config*.h" in
> our header files.  Probably would be good to eliminate that extra step
> and switch remaining LOG4CXX_HAVE's to just HAVE.  Could you take a
> look at how to do that.
>
Removing LOG4CXX_xxx in configure.in is easy. Here is the diff for 
configure.in. Should I go through the source-files and change it also?

That's good, that config*.h is not included in the headers any more. This is a 
much better approach. Should config*.h moved to src then? Then the 
include-directory looks the same after installation.


Re: missing header?

Posted by Tommi Mäkitalo <to...@epgmbh.de>.
> The build had been tweaked so the output of configure is processed to
> change "HAVE_WHATEVER" into "LOG4CXX_HAVE_WHATEVER".  That processing
> is no longer needed since we do not include any of the "config*.h" in
> our header files.  Probably would be good to eliminate that extra step
> and switch remaining LOG4CXX_HAVE's to just HAVE.  Could you take a
> look at how to do that.
>
How do you handle these LOG4CXX_HAVE_LIBXML2, LOGCXX_HAVE_ODBC and others, 
which are used in the headers?

Re: missing header?

Posted by Curt Arnold <ca...@apache.org>.
On Dec 28, 2004, at 4:07 PM, Tommi Mäkitalo wrote:


> Hi Curt,
>
> I just checked out a fresh version and realized, that you checked in a
> nonworking src/transcoder.cpp. You added a "#iinclude 
> <log4cxx/config.h>".
> Did you change config_auto.h to config.h?

Sorry, comment out the include statement for now.

In the Ant build under windows, I generate an config.h which is a 
simple include of config_msvc.h.  I don't see where I do the same thing 
for a *nix build and generate a config.h that includes config_auto.h.  
I probably should just commit a config.h that is just a check of 
various compiler identifying macros and inclusions of the appropriate 
config_XXX.h" file.

The build had been tweaked so the output of configure is processed to 
change "HAVE_WHATEVER" into "LOG4CXX_HAVE_WHATEVER".  That processing 
is no longer needed since we do not include any of the "config*.h" in 
our header files.  Probably would be good to eliminate that extra step 
and switch remaining LOG4CXX_HAVE's to just HAVE.  Could you take a 
look at how to do that.

I should be able to commit tomorrow.  I've tracked down a few more bugs 
that were causing tests to fail, Level::toLevel("FATAL") was not 
returning the write value, the non-default FileAppender constructors 
were not calling activateOptions,  the FileAppender destructor needed 
tweaking, timestamps in XMLLayout were truncated at 32-bit.