On Wed, 2023-08-30 at 17:07 +0100, Gavin Smith wrote: > Okay, in this case there doesn't seem to be any point in continuing the > program (and the exit status should be non-zero). We'll have to see if > these errors can be distinguished from mere warnings. > > I can imagine the current behaviour could cause problems if texi2any is > run by the "make" program as part of a build system and the build continues > after texi2any appears to complete successfully.
Yes, this is exactly the problem. But I would even argue that texi2any should abort immediately, and not produce any (wrong) output. > This Perl mechanism is not appropriate, I seem to recall, for > signalling errors in loading the module. Instead, you can use the > texinfo_register_init_loading_error and texinfo_register_init_loading_warning > functions to deliberately raise an error or warning. Yes, I know about these functions. But if it's not appropriate, why is texi2any still checking the return value of an init file? Wouldn't it be more consistent to remove it altogether? Jonas
signature.asc
Description: This is a digitally signed message part
