On Wed, Dec 22, 2021 at 12:33 PM Marius Schwarz <fedoradev(a)cloud-foo.de> wrote:
Hi,
for a new project hopefully coming to soon to Fedora, I like to know
the policy for really big data rpms.
The project could offer 18 language files for a voice recognition
system, which is ( unpacked ) up to 2.4 GB each and packed upto around
1.6~1.8 GB each.
+ 18 small ones ~50-60 MB each.
So round about, we are talking about 40 GB just for those language packs
just for the first release + a lot more for new updates per Fedora
version, and those packages grow constantly over time. Of course, users
do not need all of them at the same time, but they should be available.
Is this a valid scenario for the Fedoraproject or would this be a nogo?
I'm not sure if this is a good idea. For example, storage space in
koji and especially on mirrors of Fedora repositories is already quite
constrained, so adding tens of gigabytes to that (for every release +
for stable/updates/testing repos) would probably explode some things
:) Would it be possible to modify the software in question to download
these data files on demand instead?
Fabio