[resolved if not understood] [gentoo-user] new emerge failures

[Jack]

On 2022.06.10 13:01, Jack wrote:
> My usual "emerge -auDvN world" yesterdau failed over a bunch of slot
> conflict for various dev-qt packages (appeared to be related to :0/30
> vs :0/30.30).  However, in process of trying various options to at
> least get something emerged, I have ended up getting, for most (but
> not all) invocations of emerge, something like:
>
> Total: 114 packages (61 upgrades, 5 new, 48 reinstalls), Size of
> downloads: 997,745 KiB
> Conflict: 1 block (all satisfied)
> Traceback (most recent call last):
>  File "/usr/lib/python-exec/python3.9/emerge", line 51, in <module>
>    retval = emerge_main()
>  File "/usr/lib/python3.9/site-packages/_emerge/main.py", line 1328,
> in emerge_main
>    return run_action(emerge_config)
>  File "/usr/lib/python3.9/site-packages/_emerge/actions.py", line
> 3954, in run_action
>    retval = action_build(emerge_config, spinner=spinner)
>  File "/usr/lib/python3.9/site-packages/_emerge/actions.py", line
> 468, in action_build
>    mydepgraph.display_problems()
>  File "/usr/lib/python3.9/site-packages/_emerge/depgraph.py", line
> 10485, in display_problems
>    self._compute_abi_rebuild_info()
>  File "/usr/lib/python3.9/site-packages/_emerge/depgraph.py", line
> 923, in _compute_abi_rebuild_info
>    if dep.parent.installed:
> AttributeError: 'SetArg' object has no attribute 'installed'
>
> That one was for a full world update, but I've also gotten it for
> single packages.  Some single package invocations worked, such as
> updating portage itself from 3.0.30-r3 (stable) to 3.0.30-r4 (~amd64)
> but the newer version has the same failures.
>
> I've done a bit of searching, but haven't been able to find anything
> relevant.  This failure doesn't seem related to any particular
> conflict, as it appears immediately after what looks like a normal
> list of packages to be upgraded or reinstalled.
>
> Thanks for any suggestions about which fine manual I should read, or
> which version of portage might work any better.
>
Well, I finally managed to get all updates and dependencies emerged.  I  
still have no idea what caused the AttributeError, but I do have an  
idea what triggered it.  gnutls (and one or two others) needed  
upgrading, and caused lots of rebuilds.  I believe the error was  
triggered by a change in subslots, although I admit I don't fully  
understand how that works.

As the main example: "eix gnutls" (after upgrade)
[I] net-libs/gnutls
      Available versions:  3.7.2(0/30)^t 3.7.3-r1(0/30)^t  
~3.7.5(0/30.30)^t 3.7.6(0/30.30)^t {brotli +cxx dane doc examples guile  
+idn nls +openssl pkcs11 seccomp sslv2 sslv3 static-libs test test-full  
+tls-heartbeat tools valgrind verify-sig zlib zstd ABI_MIPS="n32 n64  
o32" ABI_S390="32 64" ABI_X86="32 64 x32"}
      Installed versions:  3.7.6(0/30.30)^t(11:02:20 AM 06/12/2022)(cxx  
idn nls openssl seccomp tls-heartbeat tools zlib -brotli -dane -doc  
-examples -guile -pkcs11 -sslv2 -sslv3 -static-libs -test -test-full  
-valgrind -verify-sig -zstd ABI_MIPS="-n32 -n64 -o32" ABI_S390="-32  
-64" ABI_X86="32 64 -x32")
      Homepage:            https://www.gnutls.org/
      Description:         A secure communications library implementing  
the SSL, TLS and DTLS protocols

I was uprading from 3.7.3-r1(0/30) to 3.7.6(0/30.30).  Some emerge  
attempts showed lots of slot conflicts between installed programs  
needing (0/30) and the propsed upgrade to (0/30.30).  I don't know why  
this seemed to confuse portage, but I ended up doing an "emerge -1  
--nodeps gnutls" and a later "emerge -auDvN world" got all the required  
rebuilds.

I suspect the AttributeError is a bug in portage, only triggered by  
some apparently unlikely combination of pending upgrades and rebuilds  
due to the subslot change.  Since I can no longer reliably trigger it,  
I don't plan on reporting it, unless someone thinks it's worth the  
effort.

Jack