Clive, It is great to hear that you are already interested in gtm.
The good news is that during the Debian packaging we solved the bootstrapping problem. We worked along with upstream and with Debian developers to find a suitable solution. The reason why gtm is needed in order to build gtm, is that it pre-processes a set of input files, and generates code from them. The pre-processing being done by a gtm compiler. What we did to break the self-dependency, was to include in the source distribution (with the agreement of upstream) the sources of the generated files. In this way, the build process starts from the full collection of sources (including the generated files) and no initial gtm compiler is needed anymore. In addition, to ensure consistency, once we build the gtm compiler, we do a second round in which we regenerate those files the traditional way, and do a second build pass. In this manner we ensure consistency with the build process that gtm has used so far. To instrument all this, we also created a CMake configuration for gtm, and that helped a lot to control all the details of the build. The reorganized code that we worked on with upstream is in this github repository (and branch): https://github.com/luisibanez/fis-gtm/tree/hackathonjune2012-brad The upstream team will be making a fresh release at some point with this new collection of files, at: http://sourceforge.net/projects/fis-gtm/ (not there yet). If one works from the CMake-ified sources (in the github repository above), the package can be built by using only gcc through: cmake . make -j 4 make install There were other modifications details, that we will be happy to describe further if anyone is interested. With this fis-gtm package, we were also able to build the package for VistA. (where fis-gtm is the only dependency for the vista package). Luis
-- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
