In message <OKvw6.112996$W05.21580626 / news1.rdc1.mi.home.com> jkaurin / home.com writes: > I tried your suggestions above and appear to get a clean compile for > tcltklib (at least no errors reported) during the make cycle. But Tk > still will not work. Well.... actually that is not abnormal. Make process is not obstacled by unavialble libraries for ext modules. In that case Makefile for the ext module is not created so ext module is not built. Is there a Makefile in ext/tcltklib? My previous post is a way to inform appropriate include/library path. If correct location is specified, make process can find out include files and libraries needed and so Makefile is created and the ext module will be built, if no errors occur. > It appears that "require tk" wants to load "tcltklib" and "tkutil" > but the associated Ruby files do not exist in my distribution. You have ext/tcltklib directory and its contents in your archive, don't you? Non existent file is binary module, right? Then it means you failed to pass corect paths to configure and the module was not built. ....Hmm you can install without tcltklib and later separately build it with ordinary way of installing ext modules. That is: 1. cd to the directory sources of tcltklib exists, 2. ruby extconf.rb 3. make && make install 2. will be failed due to path problem. You will get `checking for tcl.h... no' or such. In the case think giving extconf.rb appropriate options. If you think you give correct paths but cannot stop errors, executing `ruby -d extconf.rb ....' may be helpful. With `-d' extconf.rb prints all output in configuration, thus you can examine what's wrong with process. # ....Well, my guess can be completely wrong and there may be another # cause. In that time I can't help. > A side question: Where can I get descriptions of the directives you > noted above? configure --help does not list them. I'm afraid that is not documented now. Options for extconf.rb is described in README.EXT briefly but the fact that configure passes unrecognized options to ext module making process is not.... > This helped, I can now make Ruby 1.6.3. Thanks. I still get numerous > warnings, though, but they are mostly confined to two categories, > ptrmismatch and "unresolved" warnings of the type: > > ld: > Warning: Unresolved: > rb_check_type > rb_check_safe_str > ... > > The unresolved" warnings are numerous and appear throught the > make process. > > Can I do something to eliminate them? Uh, you got a runnable interpreter, then forget warnings :-) Ruby is not tested enough in 64bits environment. So more serious problems will be appeared later.... Mismatched pointer warnings may be suppressed with correct casts or appropriate type declarations. Unresolved symbol warnings are.... with them usually linker fails and can not get runnable, but you can. I'm not familiar with system `alphaev6-osf4.0f'. What I can say is functions in example are seemed to be declared in ruby.h (and may be intern.h, right?) without `extern' keywords. The compiler may complains that.... -- kjana / os.xaxon.ne.jp March 29, 2001 Time and tide wait for no man.