On Mon, Oct 26 2009, Eugene V. Lyubimkin wrote:
> Manoj Srivastava wrote:
>>
>>> But some of people-written snippets have, often doing it wrong.
>>
>> Can you point to some examples? Have you filed bug reports?
> I filed #552389, a good example (IMO) of confusing due to complexity. There
> are plenty of packages that do checks of parameters that are passed to script
> and then doing completely nothing with them. Some packages tend to think that
> 'disappear' is equal to 'remove', others think that it's equal to 'purge', so
> either first or second set is doing wrong.
>
>>>> I also think that there might be packages that take specific
>>>> action on those cases in the future; since in all cases packages are
>>>> being removed or disappearing. Having information that distinguishes
>>>> which part of the state transition is in effect is information may
>>>> be useful, and I see little benefit in removing it.
>>
>>> Can you elaborate on this? Not for convincing me, I'm just curious how
>>> the package can take different actions on removal depending on what
>>> package is the case of removal.
>>
>> What do you mean, how? Are you asking me a basic scripting
>> question? If not, I am failing to understand what you are asking.
> Of course, not. I questioned how this info can be useful, not how to
> write a code for it.
How can it be useful? Well, If I had needed to take different
action (conflicting package installed as opposed to user removal), the
information would have been critical.
If you meant *why* such distinctions are useful, then the answer
is probably that they have not been, so far. But perhaps some packages
want to handle removal differently from the package being fully
overwritten.
>
>> In any case, look at /var/lib/dpkg/info/ucf.postinst
>>
>> It gives examples of where things can be put in, were you
>> inclined to do different things based on how the script is called.
> It gives only a code template to do different things, but actually
> does nothing.
Right. It has not had to. Yet.
Anyway. I do not have much more to add on this subject, so this
is my last post on it.
manoj
--
A feature is nothing more than a bug with seniority. Unknown source
Manoj Srivastava <sriva...@debian.org> <http://www.debian.org/~srivasta/>
1024D/BF24424C print 4966 F272 D093 B493 410B 924B 21BA DABB BF24 424C
--
To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
