On 23-02-06 02:31PM, Omar Polo wrote:
> On 2023/02/07 00:08:40 +1100, Mark Jamsek <mark@jamsek.com> wrote:
> > The gotd build appears to be broken at the moment:
> > 
> > ----8<--------
> >   $ make server
> >   ...
> >   cc -O0 -g  -o gotctl gotctl.o error.o imsg.o pollfd.o sha1.o -lutil -lz -lm -levent
> >   ld: error: undefined symbol: got_object_id_hex
> >   >>> referenced by error.c:374 (/home/mark/src/got/gotctl/../lib/error.c:374)
> >   >>>               error.o:(got_error_no_obj)
> >   >>> referenced by error.c:391 (/home/mark/src/got/gotctl/../lib/error.c:391)
> >   >>>               error.o:(got_error_checksum)
> >   cc: error: linker command failed with exit code 1 (use -v to see invocation)
> >   *** Error 1 in gotctl (<bsd.prog.mk>:126 'gotctl')
> >   *** Error 2 in /home/mark/src/got (Makefile:49 'server')
> > -------->8----
> > 
> > The below diff seems to produce a clean build, with both regress/cmdline
> > and regress/gotd passing.
> 
> Oops, sorry, I forgot (again) to build everything after a change.  I
> knew that there were two got_repo_read_gitconfig implementations but
> only tweaked one :(
> 
> will pay more attention in the future; thanks for fixing the issue! :)

No worries! I only caught it by chance when I went to setup the new
mirror :)

stsp made me think of writing a script to run the entire test harness
daily to help catch things early. I've also been tempted to alias `gs`
to a script that parses the changeset and runs the corresponding
regress/cmdline scripts to make sure the modified files haven't broke
anything before sending the commit; like a pre-send hook :)

-- 
Mark Jamsek <fnc.bsdbox.org>
GPG: F2FF 13DE 6A06 C471 CA80  E6E2 2930 DC66 86EE CF68