The two functions are defined if the BSD-specific apm headers are found and
if libapm is not found. Both of these conditions will be true on BSD
systems, so we use #elif instead of the separate #if statements.
If autoconf finds an apm.h other than the one belonging to libapm, we get
builds errors like the ones reported by Haroldo Gambini Santos in [1].
[1] http://lists.windowmaker.org/dev/msg07937.html
In particular,
- We need to link against gdk_pixbuf-2.0 in addition to
gdk_pixbuf_xlib-2.0.
- The header gdk-pixbuf-xlib.h has been moved to the gdk-pixbuf-xlib
directory.
- The function gdk_pixbuf_new_from_file now requires two arguments.
- We need to define GdkColor ourselves since we don't include gdk.h.
Inspired by a patch by Nerijus Baliunas [1].
[1] http://lists.windowmaker.org/dev/msg07994.html
In particular,
- Remove option to compile with deprecated GDK-PixBuf 1.0.
- Remove *ROOT variables; libdockapp should be in the default include search
path and X11R6 directories are deprecated.
- Remove /opt/gtk2/bin from PIXBUF_CFG; pkg-config is likely elsewhere.
- Remove CPPFLAGS from CFLAGS and INCDIRS and DEFINES from CFLAGS; we don't
want these to disappear if user defines different flags.
Inspired in part by similar patches in Gentoo [1] and by Nerijus Baliunas [2].
[1] http://mirrors.telepoint.bg/gentoo-portage/x11-plugins/wmmenu/files/wmmenu-1.2-Makefile.patch
[2] http://lists.windowmaker.org/dev/msg07994.html
In particular,
- Add new --version command line option. (This was already documented in
the manpage.) Remove this information from the --help text. Also took
the opportunity to properly align the --fire line in the --help text.
- Remove WMFSM_VERSION macro in favor of PACKAGE_VERSION from autconf.
- Remove hard-coded version number at top of source file for
maintainability.
In particular,
- Apply, with some modification, a patch by Jari Aalto
<jari.aalto@poboxes.com> to clarify the format of the configuration file.
See [1].
- Escape minus signs to avoid hyphen-used-as-minus-sign Lintian warning in
the Debian package.
- Remove docbook-to-man comments to avoid manpage-has-errors-from-man
Lintian warning in the Debian package.
- Add commas between short and long options for readability.
[1] https://bugs.debian.org/cgi-bin/bugreport.cgi?archive=yes&bug=226274
The individual patches no longer seem to exist, so I submit them together.
According to debian/changelog entries by Arthur Korn <arthur@debian.org>,
these patches include:
* fixed memleak introduced in 0.34-2 with the patch by Timo Benk.
global char *mp[] was assigned the result of stdup() repeatedly
without free. My ugly fix is to free it all up at the beginning of
readFileSystems(). Now I'm a proud Perl coder ... closes: #196057
* memleak news: avoid it if there are more than 9 filesystems too.
* Fixed filesystem name corruption, thanks to Ralf Horstmann and Cristian
Ionescu-Idbohrn for the patch.
In particular,
- Update AM_INIT_AUTOMAKE to silence "two- and three-arguments forms are
deprecated" warnings.
- Update AC_DEFINEs to silence "missing template" warnings.
There is no need to have a variable holding the interface name and
doing the exercise of stripping the trailing colon. It is enough
to just compare the line to check whether it contains the device
name. And only if it does we sscanf() its contents.
After a recent change in the kernel (their commit 6e094bd) my
wireless interface name is 8 characters long, wlp3s0b1. But
the wmnet code to read the received bytes
totalbytes_in = strtoul(&buffer[7], NULL, 10);
assumed that the interface name in /proc/net/dev would be 6 characters
long (and would start reading at position 7).
In the linux kernel mailing list I reported the regression and none
other than Al Viro replied,
> I am not sure if 'wmnet' could do this better (any suggestions?),
*snort*
well, yes - it's called scanf(). And if one is really, really nervous
about the overhead of <gasp> parsing a bunch of integers (as if fopen/
fgets/fclose alone won't cost enough to make constantly calling that
sucker a bad idea), just use ptr + <something> - 6 instead of
&buffer[<something>] in there. That thing has just found where the
colon was (and replaced it with NUL), so dealing with "the first field
turned out to be too long and shifted everything past it" isn't hard.
******
I also took the oportunity to remove the unused variables diffpackets_{in,out}
