On 2014-04-07 17:11+0200 Thorsten Behrens wrote: > [...]The general cause of all trouble is within the file > CMakeDInformation.cmake (which is burried deep inside the path > \cmake\modules\language_support\cmake).
Agreed. > > As far as I understand it, this module is intended to define several CMake > variables/properties for the different environments the project might be > built under. In lines 178-204 some variables/properties regarding OS and D > compiler used are set up. *Only* if the OS is Windows and the compiler is > not GDC, then CMAKE_LINK_LIBRARY_FLAG is set to "-L+"). and possibly SET(CMAKE_LIBRARY_PATH_FLAG "-L+") causes problems as well. > [...]Something else strikes me odd with this CMAKE_LINK_LIBRARY_FLAG variable. > Most other variables being set in that block of code contain an explicit > reference to the D language ... and because of this most probably do not > interfere with other languages. Only the variables > CMAKE_LINK_LIBRARY_FLAG, CMAKE_LIBRARY_PATH_FLAG and > CMAKE_LIBRARY_PATH_TERMINATOR do not refer specifically to the D language. > Perhaps, when introducing language support for D, it was forgotten to make > these variables language dependent? No, these variables are also set in official CMake language support files as well. But the way they are set is much more generic and compiler- and platform-dependent rather than computer-language dependent. This is because CMake language support has become much more sophisticated since the early history of CMake when CMakeDInformation.cmake was implemented by a non-CMake group of developers. Furthermore, there is no mention of the -L+ flag in _any_ of the official methods where the above variables are set so either that was a mistake by the implementers of CMakeDInformation.cmake or only valid for an extremely narrow set of circumstances. For example, if http://msdn.microsoft.com/en-us/library/windows/hardware/ff552114(v=vs.85).aspx is relevant documentation for this compiler option it does appear to be specific to "control source display and program stepping options" so nothing to do with linking and thus likely a mistake for Windows platforms. > I agree with you, that it is indeed quite easy to just use a big hammer > for this small nail, but this might cause bigger problems for whoever > needs PLplot using the D language. No, this workaround does not affect D language support on Unix which by and large works without issues. It only affects those using the D language on Windows. Furthermore, to my knowledge nobody has tried that combination yet. So by definition that is an experimental combination where setting -DENABLE_d=OFF by default (but allowing the user to set -DENABLE_d=ON if they so desire) is the correct approach in my view. Thorsten (or anybody else here) if you do have access to a D compiler on Windows and want to look further into D language support on that platform, I would encourage you to experiment with Windows-only changes to our D language support files (using -DENABLE_d=ON), but otherwise I don't think it is possible to figure out what changes are needed in those files for the Windows case. Alan __________________________ Alan W. Irwin Astronomical research affiliation with Department of Physics and Astronomy, University of Victoria (astrowww.phys.uvic.ca). Programming affiliations with the FreeEOS equation-of-state implementation for stellar interiors (freeeos.sf.net); the Time Ephemerides project (timeephem.sf.net); PLplot scientific plotting software package (plplot.sf.net); the libLASi project (unifont.org/lasi); the Loads of Linux Links project (loll.sf.net); and the Linux Brochure Project (lbproject.sf.net). __________________________ Linux-powered Science __________________________ ------------------------------------------------------------------------------ Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test & Deployment Start a new project now. Try Jenkins in the cloud. http://p.sf.net/sfu/13600_Cloudbees_APR _______________________________________________ Plplot-devel mailing list Plplot-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/plplot-devel
