Source-Changes-D archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: CVS commit: src



> | > I think that vasnprintf and asnprintf are not used by anything in
> | > heimdal and can safely be removed. Combined with the winsize fix,
> | > this fixes the cygwin problems with minimal changes. I am trying a
> | > build now.
> | 
> | Note roken.h includes <resolv.h> and <arpa/nameser.h> that don't
> | exist on Cygwin so we had to handle it in src/tools/compat/configure.
> | (toolchain/29032)
> 
> Which is fine; I'd rather have one place to keep roken.h and deal with
> portability in compat, rather than 2.

But src/include/heimdal/roken.h is a generated file for NetBSD
with unusual method.

We might be able to generate roken.h for tools host from
src/crypto/dist/heimdal/lib/roken/roken.h.in using
src/crypto/dist/heimdal/lib/roken/roken.awk as defined
in src/crypto/dist/heimdal/lib/roken/Makefile.am:
---
roken.h: make-roken$(EXEEXT)
        @./make-roken$(EXEEXT) > tmp.h ;\
        if [ -f roken.h ] && cmp -s tmp.h roken.h ; then rm -f tmp.h ; \
        else rm -f roken.h; mv tmp.h roken.h; fi

make-roken.c: roken.h.in roken.awk
        $(AWK) -f $(srcdir)/roken.awk $(srcdir)/roken.h.in > make-roken.c
---
but it requires all macros like HAVE_FOO referred in roken.h.in
and we have to add checks for them into src/tools/compat/configure.ac,
as defined in src/include/heimdal/config.h configured for NetBSD.

Is it really worth than adding a manually edited dumb roken.h for tools?
---
Izumi Tsutsui


Home | Main Index | Thread Index | Old Index