tech-pkg archive

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

Re: meta-pkgs/mate build issues



On Fri, Feb 09, 2024 at 02:17:19PM +0100, triaxx%NetBSD.org@localhost wrote:
I tried to build meta-pkgs/mate on
$ uname -a

NetBSD goliath.triaxx.org 10.99.10 NetBSD 10.99.10 (GOLIATH) #0: Mon Jan 29 10:43:35 CET 2024 triaxx%goliath.triaxx.org@localhost:/usr/obj/sys/arch/amd64/compile/GOLIATH amd64

with pkgsrc branch
$ cat /usr/pkgsrc/CVS/Tag
Tpkgsrc-2023Q4

I have PKG_DEVELOPER=yes in my /etc/mk.conf

I got the following error for x11/mate-panel:
=> Checking for missing run-time search paths in mate-panel-1.26.4
ERROR: lib/mate-panel/libclock-applet.so: /usr/pkg/lib/libxml2.so.2: libxml2-2.10.4nb6 is not a runtime dependency

*** Error code 1
[...]
Same issue for editors/pluma and same fix:
=> Checking for missing run-time search paths in pluma-1.26.1nb5
ERROR: bin/pluma: /usr/pkg/lib/libxml2.so.2: libxml2-2.10.4nb6 is not a runtime dependency ERROR: lib/pluma/plugins/libdocinfo.so: /usr/pkg/lib/libxml2.so.2: libxml2-2.10.4nb6 is not a runtime dependency ERROR: lib/pluma/plugins/libfilebrowser.so: /usr/pkg/lib/libxml2.so.2: libxml2-2.10.4nb6 is not a runtime dependency ERROR: lib/pluma/plugins/libmodelines.so: /usr/pkg/lib/libxml2.so.2: libxml2-2.10.4nb6 is not a runtime dependency ERROR: lib/pluma/plugins/libsort.so: /usr/pkg/lib/libxml2.so.2: libxml2-2.10.4nb6 is not a runtime dependency ERROR: lib/pluma/plugins/libspell.so: /usr/pkg/lib/libxml2.so.2: libxml2-2.10.4nb6 is not a runtime dependency ERROR: lib/pluma/plugins/libtaglist.so: /usr/pkg/lib/libxml2.so.2: libxml2-2.10.4nb6 is not a runtime dependency ERROR: lib/pluma/plugins/libtime.so: /usr/pkg/lib/libxml2.so.2: libxml2-2.10.4nb6 is not a runtime dependency ERROR: lib/pluma/plugins/libtrailsave.so: /usr/pkg/lib/libxml2.so.2: libxml2-2.10.4nb6 is not a runtime dependency

*** Error code 1

The issue has probably never materialized in practice because various
other MATE packages explicitly depend on libxml2. Thanks for catching
it.

It's not as simple as having PKG_DEVELOPER=yes set to detect this. All
my build environments have:

PKG_DEVELOPER=		yes
CHECK_FILES_STRICT=	yes

I have never received this warning (being the person who updated these
packages multiple times in recent years). With my settings above, I
get:

=> Checking file-check results for pluma-1.26.1nb5
=> Checking for non-existent script interpreters in pluma-1.26.1nb5
=> Checking file permissions in pluma-1.26.1nb5
=> Checking for PIE in pluma-1.26.1nb5
=> Checking for RELRO in pluma-1.26.1nb5
=> Checking for missing run-time search paths in pluma-1.26.1nb5
=> Checking for work-directory references in pluma-1.26.1nb5
WARNING: [check-fakehome.mk] fake home directory not empty:
        /home/disciple/NetBSD/pkgsrc/editors/pluma/work/.home
        /home/disciple/NetBSD/pkgsrc/editors/pluma/work/.home/.local
/home/disciple/NetBSD/pkgsrc/editors/pluma/work/.home/.local/share => Creating binary package /home/disciple/NetBSD/pkgsrc/editors/pluma/work/.packages/pluma-1.26.1nb5.tgz
===> Building binary package for pluma-1.26.1nb5

A quick look through various things didn't turn up an immediate answer
for me. I changed the invocation of readelf to see what I would get
without errors suppressed.

--- check-shlibs-elf.awk.orig
+++ check-shlibs-elf.awk
@@ -102,7 +102,7 @@
 }

function checkshlib(DSO, needed, rpath, found, dso_rpath, got_rpath, nrpath) {
-       cmd = readelf " -Wd " shquote(DSO) " 2> /dev/null"
+       cmd = readelf " -Wd " shquote(DSO)
        while ((cmd | getline) > 0) {
                if ($2 == "(RPATH)" || $2 == "(RUNPATH)") {
sub("^[[:space:]]*0[xX][[:xdigit:]]+[[:space:]]+\\(RU?N?PATH\\)[[:space:]]+Library ru?n?path: \\[", "")

Then I get:

[...]
=> Checking for PIE in pluma-1.26.1nb5
=> Checking for RELRO in pluma-1.26.1nb5
=> Checking for missing run-time search paths in pluma-1.26.1nb5
readelf: Error: Not an ELF file - it has the wrong magic bytes at the start
=> Checking for work-directory references in pluma-1.26.1nb5
[...]

Clearly something here is behaving differently than expected.

Regards,

Dave


Home | Main Index | Thread Index | Old Index