On Wed, Sep 11, 2013 at 05:19:42PM +0300, Zeeshan Ali (Khattak) wrote: > So just because 'packaging system' is different, requirements become > *completely* different? I'm not at all saying the fedora's > requirements map exactly to gnome's but given that they are > practically very similar, the main difference I could see would be of > disk usage. CPU and RAM requirements ought to be very close, if not > the same. You are missing my point, I'm saying that ostree based images are very different from the GNOME live CDs that are already listed in libosinfo database, and as such, they should go to a different namespace, ie something that is not <os id="http://gnome.org/3.10";>. If someone was to make a GNOME 3.10 live CD, we want that live CD to get this id, and we will be stuck if we used it first for ostree images. > > > Moreover, if we start telling application writers that they can use the > > 'gnome3.10' OS to get information about an ostree system, then telling them > > differently at a later point would arguably be an ABI break, > > Updating and improving existing data is in no way ABI break. If thats > the case, we have been breaking apps quite a lot and I have never > heard you say anything against it before. What I mean is if we tell application developers to lookup the osinfo database entry whose id is 'http://gnome.org/3.10' to get hardware requirements for ostree images, it would not be nice to change our mind later and tell them that now they should be using 'http://ostree/3.10'. We don't have an API in libosinfo which returns the id to lookup for ostree images, so what you are suggesting is that we tell applications they can hardcode a given ID, right? Christophe
Attachment:
pgpoCRKfJZp0W.pgp
Description: PGP signature
_______________________________________________ Libosinfo mailing list Libosinfo@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libosinfo
