On Thu, Feb 08, 2001 at 09:07:11AM -0600, Richard Braakman wrote:
> On Thu, Feb 08, 2001 at 03:39:33PM +0100, Domenico Andreoli wrote:
> > the question is which dependencies should curl-ssl have?
> >
> > i'm thinking something like:
> > Conflicts: curl(<7.4.2-2)
> > Depends: curl(>=7.4.2-2), libcurl0-ssl(=7.4.2-2) | libcurl1-ssl(>=7.5-1)
>
> If you Depend on curl >= 7.4.2-2, then you don't need to also Conflict
> with older curl.
>
> > the correct relation should be this:
> > ( curl(=7.4.2-2) AND libcurl0-ssl(=7.4.2-2) ) OR ( curl(>=7.5-1) AND
>libcurl1-ssl(>=7.5-1) )
> >
> > can be this relation achieved someway?
>
> Using Boolean logic this transforms to:
>
> Depends: curl (= 7.4.2-2) | curl (>= 7.5-1),
> libcurl0-ssl (= 7.4.2-2) | curl (>= 7.5-1),
> curl (= 7.4.2-2) | libcurl1-ssl (>= 7.5-1),
> libcurl0-ssl (= 7.4.2-2) | libcurl1-ssl (>= 7.5-1)
>
this is really interesting, so let me understand more:
1) the first clause can be simplified as you said below, ok.
2&3) the second... should be "|" read as XOR (exclusive or)?
there is no meaning in the order of the operands of "|", right?
if what i said is correct then second and third cluses are ok.
4) the fourth clause has sense only if "|" has a XOR meaning.
please correct me if i'm not right
> The first clause can be simplified to "curl (>= 7.4.2-2)" if there were
> no versions in between.
in fact there are no versions in between
>
> Doesn't each version of curl Depend on the library it needs? In that
> case it simplifies to:
>
> Depends: curl (>= 7.4.2-2)
>
this doesn't work, since curl(=7.4.2-2) depends on libcurl0 | libcurl0-ssl
while curl(>=7.5-1) depends on libcurl1 | libcurl1-ssl and i need to force
the install of the ssl versions of the libcurl libraries in order to make
curl understanding something about SSL.
> > another point. potato has a 6.0-xxx series of curl packages, at that time
> > libcurl was not even a separate package so the upgrade from potato shouldn't
> > have all this mess. all this work could be a waste of time, since some
> > upgrade problems in woody are ok since it is "testing".
>
> Then I don't think it's worth the complication.
>
i realized that at the end of the message :)
anyway thanks
> Richard Braakman
Domenico Andreoli
-----[ Domenico Andreoli, aka cavok
--[ http://filibusta.crema.unimi.it/~cavok/gpgkey.asc
---[ 3A0F 2F80 F79C 678A 8936 4FEE 0677 9033 A20E BC50
PGP signature