*** Bug 32311 has been marked as a duplicate of this bug. ***

CC: (none) => joselp

*** Bug 32344 has been marked as a duplicate of this bug. ***

CC: (none) => rihoward1

AFAIK this is out of scope for atelier and domain of new great sysadmin team

Assignee: atelier-bugs => sysadmin-bugs
CC: (none) => filip.komar

CC'ing stormi and papoteur

CC: (none) => marja11, stormi-mageia, yvesbrungard

Reassigning to stormi, because he says in bug 32437 that he should be the default assignee for madb.

Assignee: sysadmin-bugs => stormi-mageia

...and now it works to set component to madb :)

CC: (none) => fri
Component: Other => madb.mageia.org

In progress!

Status: NEW => ASSIGNED

(In reply to Samuel Verschelde from comment #7)
> In progress!

Thanks for working on it!
And nice to already see "Latest stable (9)" :-)

I think we're good now.

I also updated the "Release blockers for Mga 10", "Intended for Mga 10" and "High priority for Mga 10" pages, but I don't know if anyone uses them. Two of them were two releases late. Only the release blockers one was usable.

Fixed

Resolution: (none) => FIXED
Status: ASSIGNED => RESOLVED

Thanks, Samuel,

You rock :-D

Hi Samuel

A little problem remains :
"Versions comparison" displays something very strange 

all the green lines (added in Cauldron) show old versions
(as an example Cauldron provides Firefox 0.102.9.0 
when Mageia9 provides Firefox 0.115.4.0   ???)

and a strange thing 
in Cauldron opencpn (that I maintain) is said to be version 5.6.2 (which is the Mageia8 version) 
when it is actually 5.8.4 for Mageia9 and Cauldron

CC: (none) => philippedidier

Hi Samuel,

May I suggest a trivial improvement? I think nowadays we should switch the default architecture to x86_64 instead of i586. As a majority of people are searching for x86_64 packages, that would avoid a useless first request with i586 for the Mageia servers and the waste of time for the requester.

Thank you for fixing it so far

---

I agree with suggested improvement in Comment 13

---

This search not find the packages "samba", "Samba-client", "samba-common" etc:

https://madb.mageia.org/package/list/t_search/samba/application/0/arch/x86_64

Same for Cauldron

But if i change to Mageia 8 or 7, it do.

=> Reopening

Resolution: FIXED => (none)
Status: RESOLVED => REOPENED

The same for gajim:
https://madb.mageia.org/package/list/application/0/release/9/arch/x86_64/t_search/gajim
returns no results.

madb gets the RPM information from Sophie. I can't find gajim in Sophie on Mageia 9 either, despite it is present in the distrib-coffee mirror. Same for samba.

Sophie also believes that cauldron provides firefox 102.9.0, and there's a mismatch between the SRC RPMs and the build RPMs it sees.

http://sophie.zarb.org/search?search=firefox&type=byname&deptype=P&distribution=Mageia&release=cauldron&arch=

There are visibly issues with the repo scan of both Mageia 9 and cauldron in Sophie, with packages either missing or outdated.

CCing Nanardon, who maintains Sophie.

CC: (none) => nanardon

(In reply to Davy Defaud from comment #13)
> Hi Samuel,
> 
> May I suggest a trivial improvement? I think nowadays we should switch the
> default architecture to x86_64 instead of i586. As a majority of people are
> searching for x86_64 packages, that would avoid a useless first request with
> i586 for the Mageia servers and the waste of time for the requester.

Regarding this, I fully agree, and the change is indeed rather trivial. The only drawback, but I think we can live with it, is that shared links that previously pointed at the i586 version of lists and packages will now point at the x86_64 version.

I have tried to use madb today

"Version comparison" : 
- gives always a strange behaviour comparing Mageia9 and Cauldron
- it works well comparing Mageia8 or Mageia7 to Cauldron !!!

"Packages/Applications"
- in the list for Mageia9 some applications are missing when they are present in Mageia8 or Cauldron
- searching a particular application with "search" can find it in Mageia8 Mageia7 Mageia6 and Cauldron  but not in Mageia9

If madb uses Sophie to get information there's always something wrong inside Sophie which seems having problem to create a correct database of rpms for Mageia9
when it works correctly for Mageia8 (including even the updated rpms) for Cauldron but not for Mageia9

The strangest thing is that Sophie can find for Mageia9/core/release the srpm of this particular application but not the build rpms

There's still a problem inside Sophie but not a bug in madb

Always the same problem as described in comment 12 comment 14 comment 16

I digged a little more inside Sophie
1) for Mageia9

The database of srpms for Mageia9 seems correct
but lots of rpms (whose the srpm is present) are missing in the database of rpms
for example 
one SRPM opencpn-5.4.8-2
no rpm opencpn-5.8.4-2

Strangest
 searching rpms beginning with openc Sophie find all the rpms of plugins for opencpn that are in /core/updates_testing and only them !!!
but can't find them in core/release
Sophie seems not looking for rpms in the right repositories


2) for Cauldron
Strangest : in the database of rpms of Cauldron we can find several rpms of different versions of a same program
when there's only one srpm for this program

As an example
one SRPM of 
opencpn-5.8.4-2 

but three RPMS of 
opencpn-5.6.2-5
opencpn-5.6.2-6
opencpn-5.8.4-2

In the mirrors there's only the last version

NB versions 5.6.2-5, 5.6.2-6, 5.8.4-2 were successively created inside Cauldron before Mageia9 was published but only 5.8.4-2 version of srpm and rpm was present in Cauldron at the end, the last version replacing the previous ones

Sophie seems unable to erase old versions when a new one appears in Cauldron

Hope this comment will help Nanardon to lead Sophie to the right repositories

Sorry for the strange mid-air collision
Comment 19 might not exist it was a draft !!!

Hi
No news from Sophie ?
Lots of rpms are still missing for Mageia9/core/release or Mageia9/core/updates inside Sophie
but they appear in Cauldron/core/release sometimes several times (with their successive versions) with their mga9 suffix !

Besides this the srpms for Mageia9 are all present (inside core/release and core/updates) with their correct mga9 suffix
Besides this the srpms for Cauldron have a correct mga10 suffix

It seems like Sophie is not looking for rpms in the right place for the rpms of Mageia9 and Cauldron

That's the reason why Mageia App DB gives wrong results for versions comparison
between Mageia9 and Cauldron
That's the reason why it fails when browsing for packages or applications inside Mageia9  or Cauldron (for which it finds only mga9 rpms)

I have sent an e-mail to Olivier with your findings and a link to the issue.

Hi

Sophie seems to have been partly corrected :

Sophie now functions correctly when looking for rpms inside Mageia9 
and as a consequence Mageia App Db can list all the versions of rpms present 
in Mageia9 (aka latest release) in core repo and updates repo

That's OK for
http://madb.mageia.org/package/list/application/0/release/9/arch/x86_64


BUT Sophie doesn't look for the content of the actual Cauldron and seems to keep in memory what was Cauldron before Mageia9 is published and this as 2 consequences :

1) Mageia App Db lists an obsolete version of Cauldron rpms
as an exemple : for blender it shows 3 versions 
3.3.3-2.mga9  3.3.4-2.mga9 3.3.6-mga9
http://madb.mageia.org/package/show/application/0/release/cauldron/arch/x86_64/name/blender

When in the actual Cauldron repo it's now 3.6.9-1-mga10

2) Mageia App Db displays a totally wrong comparison between Mageia9 and Cauldron :
Cauldron seems to have older versions than Mageia9
http://madb.mageia.org/package/comparison/release/9/arch/x86_64