Max Bowsher wrote: >>> On Sun, 2003-02-02 at 07:37, Igor Pechtchanski wrote: >>>> Incidentally, there are two backslash() functions defined: one in >>>> filemanip.h, and another in concat.h. This is the only function >>>> defined in concat.h, and it doesn't seem to be used anywhere >>>> anymore. Is concat.h dead? Should it be removed? Ditto for >>>> concat.cc? > >> On 2 Feb 2003, Robert Collins wrote: >>> Not sure, lets leave this for now. > > Igor Pechtchanski wrote: >> FYI, I've built setup without concat.h and concat.cc, and it seems to >> be working fine. I'm attaching a patch that removes all references >> to concat.* files (most of the size is due to the generated >> Makefile.in changes). After applying the patch, you'll have to "cvs >> rm" concat.*, sorry (I don't know how to do this with patch). > > You can't. > > Robert, I too have tested this. Note that concat.h was only included > in 1 file, and that file doesn't use the only function declared in > concat.h. > >> * Makefile.in: Regenerate. > > One thing to note: I don't know whether Igor is still using automake > 1.7.1, or if he trimmed the patch to only contain the relevant parts, > having made it with Cygwin's current automake 1.7.2. > > What is policy for setup? Should we regenerate Makefile.in (and > aclocal.m4, and any other relevent files) with the current Cygwin > release of automake, since we are changing it at all, or would you > prefer to keep it at the version currently used?
And just after pressing send, I thought: Since libgetopt++ does not store its generated files in CVS, is there any point in setup doing it? You still need autotools to bootstrap libgetopt++, so why not setup as well, since you can't build setup without it. Max.