On 26-03-2014 12:08, Martin Bähr wrote:
Excerpts from Michael K. Johnson's message of 2014-03-26 10:55:48 +0100:
If we do that, then we can't version conary independently
from Fedora as part of a platform.
can you explain what that means?
My point is that we don't want to ask for Conary to be included
in Fedora without being integrated into process.
ok, i was not even thinking about that.
i merely mean to provide conary as an rpm package, from our own server for
example:
(hypothetical instructions to install foresight could then look like so:)
download this rpm here, install it with rpm -i conary.rpm
then run conary-setup. if there are no errors, your fl:3 is ready.
But please don't start the process of contributing a Conary RPM to Fedora
without that.
as i said, that idea didn't even cross my mind. i was merely wondering about
having conary packaged in rpm format, to make installation easier.
but i meant even for fl:3, having conary as an rpm will make it easier to
convert a fedora install into foresight. there are of course limitations, as we
can't deal with additional repositories a user may have that we have not
imported.
No, it really doesn't make it much eassier.
ok, that's true, the difference between downloading a tarball and downloading
an rpm is not that big. but it makes it feel easier. if i am a fedora user,
installing an rpm gives me at least the assurance that this won't scribble
randomly all over my packages, and that i can trust rpm to erase it again if i
want to.
and i think it will make a difference when creating a respin.
I get what you're saying Martin.
However, before we consider creating a process which allows for
perceived 'risk-free' trial runs of conary adoption (which is
essentially what you are proposing IIUC), we should probably ask
ourselves whether this feature is actually worth anything in the grander
scheme of things?
After all, any sane developer would probably rather play around with the
adoption process on a non-production system such as a separate box or
even just a VM? Doubly so if we politely point out that we explicitly
point out that adoption is a currently a unidirectional process and that
we do not support the 'un-adoption' of a previously 'adopted' f20 system?
That said, the idea of having a clean way to 'un-adopt' a system might
count for something to those of a more skeptical bent, as it will show
that we ('we' he says) have taken great care to create a clean
'adoption' and 'un-adoption' process between the 'conary tier' and the
'f20 system tier'. I'm just not convinced that it is worth doing unless
it is really simple and straightforward to do with the tools already at
our disposal...
/Rune
_______________________________________________
Foresight-devel mailing list
[email protected]
https://lists.foresightlinux.org/mailman/listinfo/foresight-devel
