[Mageia-sysadm] Decision needed on mgaapplet and adding Requires.

Michael Scherer misc at zarb.org
Sat Aug 20 11:07:30 CEST 2011


Le vendredi 19 août 2011 à 17:27 +0200, Thierry Vignaud a écrit :
> On 17 August 2011 00:53, Michael Scherer <misc at zarb.org> wrote:
> >> Either mgaapplet must be changed to allow requires to be satisfied
> >>  from Core Release, or newly added requires must be copied from
> >> Core Release to Core Updates Testing, and then pushed to Core
> >> Updates, when testing has been completed.
> >>
> >> If mgaapplet is to be modified, this must be given a high priority,
> >> as it is now blocking another update.
> >> https://bugs.mageia.org/show_bug.cgi?id=2353
> >
> > I would modify mgaapplet, personnaly.
> 
> A few remarks:
> 
> 1) This will make MageiaUpdate wey slower
>    (quite a lot more synthesis (and bigger ones!!!) to parse,
>     quite a lot more of requires to parse/compute)

Can't we do it just when needed, ie, when installing the update, not
when showing them ? This would reduce the need to parse at the start all
hdlist, and would be as fast as urpmi.

> 2) Even with that behaviour, we WILL STILL FAILL on some machines
>    if user has eg DVD media but not full network media
>    and thus some packages may still be missing from DVD

That's quite unavoidable, if we add a dependencie, we cannot take in
account the fact that some people will just have the dvd. If I add a
dependencie on texlive, should we copy texlive to update ? that would be
wasteful. 
Should we take the fact that some people just have the one, and so do
not have any network source, nor dvd, and copy everything ?

And where is the tool to check that, should we place extra burdain to
sysadmin/qa/dev just for that ?

> 3) Fred Lepied, Frederic Crozat & Warly have decided to remove
>    all code that was adding full media and the like, checking for
>    missing media, ... b/c of that
> 
> 4) we wouldn't cover people not upgrading mgaonline first

That would be temporary problem, once the upgrade is done, this would no
longer cause issue. While the popup is annoying, the system of people
didn't stopped working, no data have been lost.

> 5) we know for a fact that copying new requires works smoothly
>     vs a new procedure to test

the new procedure would just be the same as "using urpmi to upgrade", so
that would reduce the number of thing to tell to people.

> As we are talking about an already product, I would go
> to the known to work way...

And not fix it properly, like it was not fixed for the previous year,
and not document anything ? 

Copying stuff to updates is error prone, take ressource for everybody by
duplicating rpm and wasting time to generate hdlist ( because each rpm
needlessly take space and cpu time for hdlist ).

> I don't think that's the right moment to reinvent a new way to
> do updates when we already know one.

Sure, so when is the right moment ? Given that no one cared for the
previous years, I would say that "no moment is the right one", despites
clear problems. So this moment seems as good as the others.

For pratical reasons, I would concede to use the copy method for mageia
1, temporary. And to fix for mageia 2, without exception, if no one fix
mga update in the 8 remaining months we have, then we have a bigger
problem.

-- 
Michael Scherer



More information about the Mageia-sysadm mailing list