| Summary: | dev-db/pgmodeler-1.1.5, 9999 ebuilds | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | CaptainBlood <luc_pierard_de_maujouy> |
| Component: | Current packages | Assignee: | PgSQL Bugs <pgsql-bugs> |
| Status: | UNCONFIRMED --- | ||
| Severity: | normal | CC: | eschwartz, louis.frayser, luc_pierard_de_maujouy, patrick, pgsql-bugs, ThyrusG, titanofold |
| Priority: | Normal | ||
| Version: | unspecified | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Attachments: |
tar directory
tar directory tar gz directory |
||
amd64 here. Thanks for looking into this. Recommendation: you can avoid redefining src_prepare() by moving xml-error.patch into a global PATCHES=() variable. The commented out older version of DEPEND could also be dropped since it's still able to be seen in the old ebuild. ;) Moving to SLOT=6 is an absolute dealbreaker, it cannot be done -- this says that pgmodeler 0.9.4 and 1.1.5 can be installed at the same time as each other, but they install the same files... what's the purpose of that change? Agree slotting albeit reflecting qt level should be cancelled. User should be offered to alternatively install a qt5 XOR qt6 version. I know PATCHES usage is recommended. This is from a PERSONAL overlay, sometimes infringing gentoo packager tendencies as a whole. Removal of comments could be considered at least level of priority, when not explicitly stated as a rule somewhere. It has no practical impact on the final user experience. Thks 4 the advice however. Tested the 1.1.5 ebuild I changed the SLOT back to "0" and emerged pgmodeler-1.1.5. The emerge pulled in postgresql-14, so I tested with that version of the database. I was able to connect with pgmodeler and view my old tables*. -- *I currently use posgresql-15. (In reply to CaptainBlood from comment #4) > Removal of comments could be considered at least level of priority, when not > explicitly stated as a rule somewhere. It matters because it's not clear if your submission is complete or if it's a TODO, or if those things need addressing still, or what. (In reply to CaptainBlood from comment #3) > I know PATCHES usage is recommended. > This is from a PERSONAL overlay, sometimes infringing gentoo packager > tendencies as a whole. I understand, but you're submitting the ebuild here - which is appreciated - and receiving feedback as a result. All this reply really says is "I agree". Strange result for the 9999 ebuild
I changed the SLOT back to "0" and emerged pgmodeler-9999.
I installed then restored my data to postgresql-17 and tested with that version of the database and POSTGRES_TARGETS.
When emerging: emerge -av pgmodeler with POSTGRES_TARGETS=posrgresql15 potgresql17" (make.conf), I get this...
root@Ghost-of-Gentoo /usr/local/portage/dev-db/pgmodeler
# emerge --nodeps -g n -k n -av dev-db/pgmodeler
These are the packages that would be merged, in order:
[ebuild R ~] dev-db/pgmodeler-9999::local POSTGRES_TARGETS="postgres15 -postgres14 -postgres16" 0 KiB
NOTE: There's no mention of postgres17.
I even tried USE=postgres_targets_postgres17 to no avail. I checked emerge --info and 17 was there along with 15.
I was still able to connect to posrgresql-17 with pgmodeler and view my tables.
Eix pgmodeler shows...
Installed versions: 9999*l{gpkg}[1](17:19:13 11/27/24)(POSTGRES_TARGETS="postgres15 -postgres14 -postgres16"
I did grep thru /var/db/repos/gentoo/eclass for POSTGRES_TARGETS and didn't see anything suspicious.
--
Any ideas?
You might be being affected by: profiles/base/use.stable.mask:9:postgres_targets_postgres17 but it's a bit odd that you're hitting that with package.accept_keywords (as shown in your paste by the ~) which I think bypasses that. (In reply to Sam James from comment #9) > You might be being affected by: > profiles/base/use.stable.mask:9:postgres_targets_postgres17 > but it's a bit odd that you're hitting that with package.accept_keywords (as > shown in your paste by the ~) which I think bypasses that. Thanks. I found the problem: Back in March, I copied a postgres.class from the gentoo repo to the my local one. It had an old definition of _POSTGRES_ALL_VERSIONS. I just deleted it and all is well. Created attachment 912690 [details]
tar directory
slot 6 reverted to slot 0
last tar inconsistent... in progress Created attachment 912698 [details]
tar gz directory
cleanup
postgresql:17 for all as upstream seems to allow it.
Please don't tar.gz that stuff unecessarily. Simply attach unified diffs for ebuild changes, and any patches applied in FILESDIR as plaintext patch attachment. I just tried this with pgmodeler-1.1.6, which has been released in the meantime. Is seems to build just fine, except that when starting I was greeted with the notification that a new version is available, despite the NO_UPDATE_CHECK=1 in the eqmake6 call. I'm not sure how essential is to track that down though. I would be happy if a 1.1.x version would end up in the main repository. |
Created attachment 911919 [details] tar directory Only tested with gcc:14+glibc. patch in files currently commented out in ebuild, as not required when build as above. The core of the upgrade is there. Dependencies version check might be advisable, to match upstream requirements. My apologizes for not PR on github.