Jens Rieks <[EMAIL PROTECTED]> wrote: > Hi, > On Wednesday 07 April 2004 08:50, Leopold Toetsch wrote: >> - runtime/parrot/*/ > This directory does only include generated files, doesn't it?
Almost. We have in CVS: - runtime/parrot/include/DWIM.imc - runtime/parrot/include/Getopt_Long.imc - runtime/parrot/chartypes/*.TXT - runtime/parrot/include/*.pasm is autogenerated >> - library/ > Maybe move it to runtime/library? or runtime/parrot/lib > We can also create a runtime/t/ (or t/runtime) where the library tests > can go to. or runtime/parrot/t >> 1) We should probably decide to use just one for simplicity. > ack >> 2) Library search paths and appropriate helper functions inside Parrot >> are still needed. > What about an PMC array? Each PMC can then represent a directory Yep. > Parrot does the C equivalent of "set file, dir[name]", if the file returned > is a ParrotIO it is used to read the library file. This still needs platform support for e.g. path_concat(path, file) >> 3) we need some scheme to organize lib paths > What about > runtime/parrot for parrot's generated files Generated or not doesn't matter. Both must be present at runtime. > directories like runtime/perl6, runtime/tcl... Yep. >> 4) Library testing is needed as well > indeed. > is it okay to use (e.g. the streams) examples as test basis? Sure. Test is test. But more low-level stuff is needed to. Examples start with to much of functionality normally. > jens leo
