RE: [Log4cpp-devel] Rolling File Appender v2

[Bastiaan B. <Bas...@li...>]

Hi,

Alex, thanks for the contribution!
I've committed RollingFileAppender to CVS. Made a slight change to get
it to compile on Linux:=20
modified the lines with strtup() to compare using strcasecmp(), because
strtup() is not portable. Yes, neither is strcasecmp() but (at least?)
Win32 has an equivalent in the form of stricmp(). This will be fixed in
Portability.hh. After that the Win32 project files can be updated and
the SimpleConfigurator patch applied.
Also the class needs (Doxygen) documentation, paritcularly since its use
is limited to non forking programs. And until log4cpp has proper
multithreading support limited to single threading programs as well.=20
To be honest, on *nix systems I think it's much simpler to let
logrotated handle the rolling of log files. Then the daemon only needs
to let the Appenders reopen the logfiles upon recieving a HUP signal.

  =20
> -----Original Message-----
> From: Evert Haasdijk [mailto:ev...@zu...]
> Sent: Thursday, January 10, 2002 9:16 AM
> To: Alex Tapaccos; log...@li...
> Subject: RE: [Log4cpp-devel] Rolling File Appender v2
>=20
>=20
> Alex,
>=20
> Maybe I'm picking nits here, but in log4j the RollingFileAppender is
> different: it just rolls over when it reaches a certain=20
> maximum size. What
> you've made looks more like their DailyFileAppender. Maybe=20
> you should rename
> it?
>

Hmm, yes, apparently the log4j folks associate rolling primarily with
size whereas Alex and I first think of time. Therefore I already
committed the Appender before noticing the different name in log4j. I do
agree however that it's better to have the same names in both log4j and
log4cpp to avoid confusion. I'll rename it....
=20
=20
> Otherwise, it compiles with Borland C++ Builder - if you have a test
> program, I'll test that it actually works with the Builder, too.
>

Yes, a small test program for in $topdir/tests would be nice.=20

Thanks,

Bastiaan
=20
> Ta, Evert