Proposal to reduce anti-bundling requirements

[Zdenek Kabelac]

Dne 11.9.2015 v 18:22 Reindl Harald napsal(a):
>
>
> Am 11.09.2015 um 18:17 schrieb Zdenek Kabelac:
>> Dne 11.9.2015 v 16:44 Reindl Harald napsal(a):
>>>
>>> Am 11.09.2015 um 16:31 schrieb Zdenek Kabelac:
>>>> Dne 11.9.2015 v 15:47 Reindl Harald napsal(a):
>>>>> don't tell me rpmfusion could not easily make that fully automated
>>>>>
>>>>>> This Fedora plan simply puts too much work at everyone's hands.
>>>>>>
>>>>>> Sure - people who care about safety might have some option - like  I
>>>>>> always want to have ONLY the latest lib - and drop everything else,
>>>>>> but
>>>>>> there are still lot of users who could live with   older libs quite
>>>>>> happilly  (and especially in the case they do not use the library in
>>>>>> question AT ALL - which is the maint point here)
>>>>>
>>>>> you said "every one has tons of free time" - well - and who would
>>>>> maintain the
>>>>> dozen of versions of libraries packages?
>>>>
>>>> You miss few important points:
>>>>
>>>> 1.)
>>>> If you have  lib.so.2  and lib.so.4 - it may need far more work then
>>>> just running  rpmbuild   - so far away from 'fully automated'.
>>>
>>> do we now build distributions backed by "may" and "possible"
>>>
>>> in most cases it is just a "rpmbuild" like the mass-build on koji and the
>>> exceptions need some handwork, no big drama
>>>
>>>> 2.)
>>>> What maintaining time are we talking about - since Fedora breaks working
>>>> thing in the first place for no good reason and force massive
>>>> maintenance time on every user of new library in 'short' time for some
>>>> potential 'security' fixes - but you may on the other hand put in dozed
>>>> of new security breaks anyway  - and when I see how frequently i.e. gtk
>>>> libs  may break whole distro -  it would be far more pleasant to see
>>>> just couple broken apps at time - instead of rendering whole  rawhide
>>>> unusable....
>>>
>>> just because you don't understand or agree with the reasons don't mean
>>> they
>>> are not good
>>
>> In real-world developing & testing app to work with all components takes
>> significantly more time  then in Fedora 'garden'.
>
> don't explain me the real world after 12 years of software development
>
>> Not sure why you fail to understand this - I do like to have some apps
>> to be latest/greatest - while others might be rather 'tested & stable'
>> or even 'have them' (since they are no longer developed).
>>
>> More complex projects even fail to fit in 1/2 year release cycle.
>>
>> In non-Fedora 'world' it's the user who picks what he want to use,
>> however in Fedora 'garden' a few people selects what user may use and
>> puts huge walls and pointless obstacle all around if they want to use
>> something else -  yes I fail to see why this should be good for me....
>
> really?
>
> that must be the reason why i run Fedora successful for 7 years now on all
> sort of production and development servers and chose for myself which version
> of mysql, php, postfix, dovecot, dbmail and what not is running and when it is
> deplyoed independent of Fedora reelase cycles in both directions (holding
> major upgrades back as well as make them long before Fedora)
>

We are finally getting to the point....

How many machines do you need to use for that setup ?
I prefer to use 1 system on 1 hardware on 1 disk - no kvm, no qemu.
And even containers are not a good fit - thought getting close....

Why I cannot use multiple different versions of php on a SINGLE machine ?
Why I cannot use some  10 years old graphical program (unless I do a local 
static compilation, so I'm sure it will work) ?

It's called free of choice....

Zdenek

PS: And my Fedora is even older continually upgraded rawhide...