Re: Possible regression in "make all-depends-list"
Date: Fri, 03 Mar 2023 16:13:09 UTC
> On 2. Mar 2023, at 18:07, Jan Martin Mikkelsen <janm@transactionware.com> wrote: > > On 2. Mar 2023, at 17:38, Gleb Popov <arrowd@freebsd.org> wrote: >> >> On Thu, Mar 2, 2023 at 6:35 PM Jan Martin Mikkelsen >> <janm@transactionware.com> wrote: >>> >>> Running “make all-depends-list” on the hs-pandoc port shows none of the dependencies listed in the USE_CABAL variable in the Makefile. In the 2022Q2 branch, these dependencies were included. >> >> USE_CABAL dependencies aren't port dependencies. These deps are >> compiled statically into the port, so from Ports Framework perspective >> they are part of source files that make up the software being built. > > Thanks, I misdiagnosed that one. The problem was some files were discarded because of a bad checksum due to a Unicode translation issue. (Byte order marks added to a UTF8 file.) > >>> Running “make all-depends-list” on x11-wm/xfce4 shows x11-servers/xorg-server as a dependency, but it does not include all of the dependencies for xorg-server. For example: >>> >>> xfce4 $ echo $PORTSDIR >>> /data/scratch/janm/p4/ports-2023.1/ports >>> xfce4 $ make all-depends-list | egrep 'xorg-server|libxcvt' >>> /data/scratch/janm/p4/ports-2023.1/ports/x11-servers/xorg-server >>> xfce4 $ cd ../../x11-servers/xorg-server >>> xorg-server $ make all-depends-list | egrep 'xorg-server|libxcvt' >>> /data/scratch/janm/p4/ports-2023.1/ports/x11/libxcvt >>> >>> In the 2022Q2 branch the dependencies were included. > > This case still seems to be a regression, as far as I can see. I tracked down the problem with the missing dependencies. This fixes it for me. --- a/ports/Mk/Scripts/depends-list.sh 2023-03-02 11:46:56.000000000 1100 +++ b/ports/Mk/Scripts/depends-list.sh 2023-03-02 11:46:56.000000000 1100 @@ -76,7 +76,7 @@ ;; esac if [ -f ${d}/Makefile ]; then - if [ -n $f ]; then + if [ -n "$f" ]; then export FLAVOR=$f fi break Should I open a bug? Regards, Jan M.