On Fri, Dec 16, 2011 at 6:47 PM, Marcelo Vanzin <mvan...@vmware.com> wrote: > Hi Natanael, > > On 12/15/2011 05:29 AM, Natanael Copa wrote: >> I'm looking at the issue with building on Alpine Linux (uclibc): >> https://sourceforge.net/tracker/?func=detail&aid=3459926&group_id=204462&atid=989708 >> >> I wonder if it wouldn't be better to use AC_CONFIG_HEADERS and from >> configure script check for the features and then: > > I'm not sure how AC_CONFIG_HEADERS has anything to do with the problem you're > describing in the bug.
I has nothing to do with it really. I was looking at the code and felt an urgent need to clean up things since this is not first time i get a build failure with uClibc. I tend to like fix things properly rather than just add another quick and dirty fix. I was actully thinking of the getloadavg since my current patch for that no longer applied. Looking at the changelog and diffing the source (any public git/svn/mercural?) there was added yet another bunnch if #ifdefs.... > It seems that, in your case, you need to change this > check in lib/stubs/Makefile.am to not ignore Linux: > > if !LINUX > libStubsCS_la_SOURCES += stub-msgfmt-fbsd.c > endif ah... that explains. very helpful thanks! >> Are there any particular reasons that AC_CONFIG_HEADERS is not used? > > The source code in open-vm-tools is shared with lots of other VMware products. > Our team doesn't own a lot of that code, and autoconf is not generally used to > build our products. So the checks in the code are not generally based on > autoconf-style variables. I'm no big fan of autotools myself really but one of the stronger points is that they help with portability. The code seems to have checks for glibc, different versions of glibc, sun, uclibc, linux, apple (darwin?) and android and have various code paths depending on which system it is. By using autotools "properly" you could have checked for a particular feature by test compiling it in the configure script (thats what autotools normally do) and then use HAVE_SOMEFEATURE rather than keeping track of (or assume) what system that has support for that feature or not. In this parituclar case I'd prefer a if !HAVE_MSGFMT libStubsCS_la_SOURCES += stub-msgfmt-fbsd.c endif than if !LINUX || __UCLIBC__ libStubsCS_la_SOURCES += stub-msgfmt-fbsd.c endif or similar. I would like to do the same thing with the geloadavg() test, but that would use of AC_CONFIG_HEADERS and a config.h (or autoconf.h to not clash with your current config.h thats used for other purposes) SO my question is basically: woudl you be interested in some help cleaning up parts of the configure.ac and #ifdefs? I would initally aim for getting rid of all #if defined(__UCLIBC__) to minimize the risk that future releases does not fail to build due to some missing #ifdef. -- Natanael Copa ------------------------------------------------------------------------------ Learn Windows Azure Live! Tuesday, Dec 13, 2011 Microsoft is holding a special Learn Windows Azure training event for developers. It will provide a great way to learn Windows Azure and what it provides. You can attend the event by watching it streamed LIVE online. Learn more at http://p.sf.net/sfu/ms-windowsazure _______________________________________________ open-vm-tools-devel mailing list open-vm-tools-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/open-vm-tools-devel
