|
[Rivet] PDF manual build [was: installation]Andy Buckley andy.buckley at cern.chMon Jan 27 16:21:26 GMT 2014
On 27/01/14 17:09, Frank Siegert wrote: > On 27 January 2014 13:57, Andy Buckley <andy.buckley at cern.ch> wrote: >> Turning off the PDF manual build by default is possible, and perhaps >> would be welcome: what do others think? > > Ok, I've taken the liberty to commit that. Feel free to revert, should > there be any objections. Ok, thanks. Haven't checked it yet, but should be simple! I think it's worth keeping a PDF in the tarball if the manual compilation was enabled, so we can distribute a version that comes with the docs built-in. >> The build system / scripts in the doc directory try to bootstrap their >> environment to only use the uninstalled libraries from inside the Rivet >> build dir. > > I assume you are talking about these lines here: > LD_LIBRARY_PATH=$(YAML_CPPLIBPATH):$(FASTJETLIBPATH):$(HEPMCLIBPATH):$(YODALIBPATH):$(LD_LIBRARY_PATH) > \ > DYLD_LIBRARY_PATH=$(YAML_CPPLIBPATH):$(FASTJETLIBPATH):$(HEPMCLIBPATH):$(YODALIBPATH):$(DYLD_LIBRARY_PATH) > \ > PYTHONPATH=$(YODA_PYTHONPATH):$(PYTHONPATH) > > (https://rivet.hepforge.org/trac/browser/doc/Makefile.am#L16) > > But these don't point to the *un*installed libraries, do they? It used to be done that way, but finding the libs turned out to be neater in the Python code of mk-analysis-latex and mk-analysis-html, which was already needing to do something similar for the Python modules which was all but impossible to do in pure automake. See this: https://rivet.hepforge.org/trac/browser/doc/mk-analysis-latex#L19 Do you think line 23 (and the equivalent in mk-analysis-html) should be *pre*pending the discovered Rivet lib path? It didn't make a difference on my system, but I can believe that it would be more correct. I'm working on the new yaml-cpp bundling right now. Thanks, Andy PS. I don't think I ever mentioned it, but a new feature is visible on L38 and L39 of that script: if you end a path variable like RIVET_ANALYSIS_PATH with a "::" double separator, then no default "system" path will be auto-appended. This was the neatest way that I could think of to avoid the warnings that you sometimes get about finding duplicate analyses, esp. during manual building if you also have an installed Rivet in a standard location. -- Dr Andy Buckley, Royal Society University Research Fellow Particle Physics Expt Group, University of Glasgow / PH Dept, CERN
More information about the Rivet mailing list |