Mailinglist Archive: opensuse-project (169 mails)

[Maciej Pilichowski <macias@xxxxxxxxxxxxxxxx>]

Hello,

  Cristian, thank you for your comments.

> > a) pay or don't say anything -- it is just a excuse
> no, it is just a basic thing on software economics

No, in this context it is an excuse. I already have installed OS 10.3 
so if I buy it it will not differ -- for good or bad. To buy a 
software I have to be sure:
a) company is entirely responsible for what it sells
b) I can trust the development in longer perspective
c) it is not some kind of "scheme"

> and if in case 
> you had a tiny idea about the amount of work needed to do what you
> suggest you would agree.

Is there a reason that you insult me? From opensuse perspective I am a 
just a random user, but you don't know me -- so why do you say things 
like that?

I work as a developer, for several years in financial IT sector and 
one thing is solid there -- no matter how fancy new version is it is 
useless if it does not work with current infrastructure 
(software/hardware/system). Opensuse is not Suse enterprise, ok, but 
why not make it more reliable?

> it is very simple, the effort must be 
> worth, because such thing will require thousands and thousands of
> man-hours be done and hence increase the cost of the product, not
> to mention that with all posible combinations of software packages
> is almost **impossible** to ensure that it will work!!

Cristian, I agree with you. However there is a difference:
a) idea is bad, because it requires a lot of effort
b) idea is good, but it requires a lot of effort

ad.a) end of discussion
ad.b) then we can think how to make things simpler

> > Support -- of course not. I am asking for providing such feature
> > as sensitive upgrade (i.e. do not upgrade downgraded packages).
> > Ensure -- user risk.
>
> THink on the following :
>
> a) will this feature be widely used so it is a nice thing for most
> users  ?

It depends how good OS next version will be. If it works there is no 
need to looking back. But if not...

As an example -- Opensuse 10.3, let's say I downloaded it, installed 
(just from the disc) and:
1) I cannot connect to samba servers
2) I cannot print anything
3) I assume I cannot work since sax crashes while detecting my video 
card
4) I cannot access my encrypted data

(1), (2), and (4) is strictly related to backward compatibility. And I 
think that every user would like to:
a) connect to servers just like she/he used to
b) print just like she/he used to
c) access her/his data like she/he used to

  It is not "could the titlebar be more orange, I like this color", 
but rather more basic thing -- using your own computer. I can work 
because I downgraded packages all related to those issues.

> b) downgraded package depends on X , Y , Z library so 
> those depend on A,B,C...Z so we have to keep them installed as
> well.. what about if the ABI of certain Key component changes ? how
> you ensure that will ever work ? what happends if f.e B comes from
> some third party repository or if C is no longer available and the
> functionality is now provided by "F" ? just use your imagination
> and create more combinations ok ? ;-)
> It is a no-win situation, it will break and leave a non working
> package in one way or another.

There is a winning strategy here, just take a look. Opensuse 10.2 -- I 
cannot print (because of cups). So I downgraded it manually, breaking 
all dependencies. Result -- yast does not work with downgraded cups 
but I can print.

Now, opensuse 10.3 upgrade -- what situation is here? There is a 
package which is known to be downgraded, dependencies are broken.
a) should new version be installed
b) should current version be kept

?

ad.a) what for, really? this is done right now. Result, all 
dependencies are ok, but I cannot print
ad.b) dependencies are broken exactly like they were broken before (so 
no extra harm is done). Result -- I still can print

I am not saying the algorithm for improving upgrade is trivial, but is 
doable and user would benefit from it.

And speaking of manhours, which you mentioned before -- you (suse) get 
massive amount of work for free -- from developers, artists, testers, 
etc. volunteers in short, so instead of summing up the costs on your 
side, it is better to look how people can contribute again (*). But 
first of all -- there must be a will to do the change.

Thanks for your time, have a nice day
bye

(*) no, patches are not everything :-))
-- 
Maciej Pilichowski
---------------------------------------------------------------------
To unsubscribe, e-mail: opensuse-project+unsubscribe@xxxxxxxxxxxx
For additional commands, e-mail: opensuse-project+help@xxxxxxxxxxxx