Hello Simon, * Simon Josefsson wrote on Fri, Sep 22, 2006 at 02:47:20PM CEST: > Ralf Wildenhues <[EMAIL PROTECTED]> writes: > > > - Failures that stem from the incomplete HAVE_CONFIG_H transition, > > including but certainly not limited to these files: > > base64.c, argp-help.c, xvasprintf.c, check-version.c, hmac-md5.c, > > memxor.c, hmac-sha1.c > > What do you mean here? Do you mean > > -#ifdef HAVE_CONFIG_H > # include <config.h> > -#endif > > or something else?
Yes. > Have we decided that gnulib files don't need HAVE_CONFIG_H checks? If > so, maybe something should be added to the gnulib manual... IIRC there has been no decision to do so. Just a new module config-h which you can use. The only point of this was that in the ALL test (which tests almost all modules together), obviously the config-h module is used, so -DHAVE_CONFIG_H does not get passed, so the modules that guard config.h inclusion do break. I did not mean to imply that the files above need to be "fixed" in any way. Sorry, I stated that wrong in my previous post. All I meant to imply was that with the megatest/ALL directory, you have to manually add -DHAVE_CONFIG_H to the DEFS variable in the Makefile to get everything to build. Maybe there should be a way to automatize the "disable the config-h module as long as one of the modules this package uses still guards config.h"; or simpler: exempt the config-h module from the ALL test for now. Cheers, Ralf
