Summary: | catalyst 1.09 livecd-stages do not support stacked profiles | ||
---|---|---|---|
Product: | Gentoo Hosted Projects | Reporter: | Hanno Zysik (geki) <h.mth> |
Component: | Catalyst | Assignee: | Gentoo Release Team <releng> |
Status: | RESOLVED FIXED | ||
Severity: | normal | CC: | morfic |
Priority: | High | ||
Version: | unspecified | ||
Hardware: | x86 | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Attachments: | livecd-stage2 |
Description
Hanno Zysik (geki)
2004-08-31 16:18:30 UTC
this looks like a portage bug actually ... that '!!!' error output comes from portage, not catalyst that said, i've never made a livecd myself, and i hate genkernel so i wont be testing this anytime soon :p what version of portage was in that chroot ? portage version is 2.0.50-r10 also if i use flat profiles it works... also if 2004.3 will be done with stacked profiles via catalyst it will break there ;) PS: sorry that i add several times but ppl are nerving me over here and i cant concentrate :( args one more thing, i hate genkernel too - is there a way to build livecds without genkernel? ;) attach your spec file then There seems to be unresolvable bugs in genkernel. I suggest that it's support be dropped from catalyst as the default method. If not it should and needs to be made optional as a dep to catalyst, perhaps on the 'cdr' USE flag. Also add a USE = diet flag to catalyst (and default to off) to inhibit the building of dietlibc. Created attachment 38762 [details]
livecd-stage2
isnt it possible to patch the catalyst livecd-stage1/2 buildings like stage1/2/3 in bug #58840 (http://bugs.gentoo.org/show_bug.cgi?id=58840) i think it is related to that bug, isnt it? And one other things related to stacked/caseded profiles - using /usr/portage/scripts/bootstrap-cascade.sh it leaves this message but continues (on stage2): /usr/portage/scripts/bootstrap-cascade.sh: line 160: myLIBSTDC++_V3=sys-libs/libstdc++-v3: No such file or directory if anyone corrects that ;) toolchain please see the comment #10 in this bug. Notice how libstd++ messes up basic operations? It also has no place in a basic bootstrap. This is another valid reason why it should be a dep when only !build? ( libstd++ ) Please fix this. on stage3 it says it can not find /etc/make.profiles/make.defaults because it just looks into /usr/portage/profiles/default-linux/x86/2004.2/gcc34/2.6/ :) so FEATURES, USE and ARCH and may be other variables are not read :( another little bug with stacked profiles... it should look into /usr/portage/profiles/default-linux/x86/2004.2/make.defaults or on upper- or subtrees... better subtrees first and then going up one directory until it found one... although this warning occurs it continues... Well that is a bug the profile then. Please use a profile that works, and file a bug with qa@gentoo about that one. Every profile that you link to _MUST_ contain a make.defaults. release@ this bug is probably INVALID or other.. Not your bug (maybe?). i solved that inital error i hope ;) using portage 2.0.50-r10 with stacked profile default-linux/x86/2004.2/gcc34/2.6 copying a make.defaults from where-ever (or creating one) to gcc34/2.6 u get this: !!! 'str' object has no attribute 'insert' !!! 'rm -Rf /usr/portage/profiles; emerge sync' may fix this. If it does !!! not then please report this to bugs.gentoo.org and, if possible, a dev !!! on #gentoo (irc.freenode.org) with 'env-update' and 'emerge' deleting that make.defaults in 2.6 let this error disappear updating portage to 2.0.51_pre20 - then it complains about missing make.defaults, copying one to 2.6 all fine... so i hope this initial bug is fixed like that ;) couldnt test catalyst livecd-stage2 because it errors here on livecd-stage1 on pciutils merge. this happens on catalyst chroot: gcc setpci.o common.o lib/libpci.a -o setpci gcc pcimodules.o common.o lib/libpci.a -o pcimodules rm -f libpci.a ar rcs libpci.a access.o generic.o dump.o names.o filter.o sysfs.o proc.o i386-ports.o /usr/lib/gcc/i686-pc-linux-gnu/3.4.1/../../../../i686-pc-linux-gnu/bin/ld: lib/libpci.a: No such file: Datei oder Verzeichnis nicht gefunden collect2: ld gab 1 als Ende-Status zur i solved that inital error i hope ;) using portage 2.0.50-r10 with stacked profile default-linux/x86/2004.2/gcc34/2.6 copying a make.defaults from where-ever (or creating one) to gcc34/2.6 u get this: !!! 'str' object has no attribute 'insert' !!! 'rm -Rf /usr/portage/profiles; emerge sync' may fix this. If it does !!! not then please report this to bugs.gentoo.org and, if possible, a dev !!! on #gentoo (irc.freenode.org) with 'env-update' and 'emerge' deleting that make.defaults in 2.6 let this error disappear updating portage to 2.0.51_pre20 - then it complains about missing make.defaults, copying one to 2.6 all fine... so i hope this initial bug is fixed like that ;) couldnt test catalyst livecd-stage2 because it errors here on livecd-stage1 on pciutils merge. this happens on catalyst chroot: gcc setpci.o common.o lib/libpci.a -o setpci gcc pcimodules.o common.o lib/libpci.a -o pcimodules rm -f libpci.a ar rcs libpci.a access.o generic.o dump.o names.o filter.o sysfs.o proc.o i386-ports.o /usr/lib/gcc/i686-pc-linux-gnu/3.4.1/../../../../i686-pc-linux-gnu/bin/ld: lib/libpci.a: No such file: Datei oder Verzeichnis nicht gefunden collect2: ld gab 1 als Ende-Status zurück make: *** [pcimodules] Error 1 make: *** Waiting for unfinished jobs.... ranlib libpci.a make[1]: Leaving directory `/var/tmp/portage/pciutils-2.1.11-r2/work/pciutils-2.1.11/lib' !!! ERROR: sys-apps/pciutils-2.1.11-r2 failed. !!! Function src_compile, Line 53, Exitcode 2 !!! emake failed on my normal system it does this: gcc lspci.o common.o lib/libpci.a -o lspci gcc setpci.o common.o lib/libpci.a -o setpci gcc pcimodules.o common.o lib/libpci.a -o pcimodules >>> Test phase [not enabled]: sys-apps/pciutils-2.1.11-r2 >>> Install pciutils-2.1.11-r2 into /var/tmp/portage/pciutils-2.1.11-r2/image/ category sys-apps >>> dosbin: making update-pciids executable... why is this?! why does it do 'rm -f libpci.a' in catalyst chroot? anyone may explain this? ;) that pciutils problem is local and this bug will be fixed with portage 2.0.51 so this bug can be closed thanks for the time spending on this :) So I am interpreting this as dev-portage needs not care about this (non-)bug. If there is an issue with that 'str/insert' output, then please hack it to run emerge with -d and post the full traceback. default-linux/x86/2004.2/gcc34 and default-linux/x86/2004.2/gcc34/2.6 both have use.defaults and make.defaults please wait 35 minutes, sync, test with portage 2.0.51 and report back, thanks this might be related to bug #62436, try the patch there and let me know if it fixes it to comment #16 i do not need to hack portage 2.0.50. i will use 2.0.51 which works well. if you guys want this bug fixed with 2.0.50 i leave it to you to debug. to comment #17 i did my own make.defaults without oss and X and some other useless stuff and it works just fine with portage 2.0.51. to comment #18 bug #62436 is not related to this bug. bug #62436 is a bug of genkernel compiling more than one kernel, here i do not even get genkernel merged. is this still no longer a release team problem? *** Bug 62442 has been marked as a duplicate of this bug. *** closing since this is a portage issue this bug has become Bug 63400 ; take comments there |