-1, I think we should be aiming at installation time rather than size (of course without getting too extreme).

Yeah, -1 too. I think this change is getting the priorities wrong. I'd need some better explanation why the size difference makes a difference to be convinced.
(Also, I think we shouldn't crank up the compression with xz and then change it to zstd for the next release.)

-1 for the same reasons as above, based on the feedback by not only Fedora QA people.

-1 as well. This is weird and wrong, and I'm rather surprised we're going for higher xz, since that makes everything much slower.

-1

This is a recurring topic and it was always a thing working on the installer. The solution presented is not a long term solution. If install image size is a concern, the focus needs to be on minimization efforts and making sure every package owner who provides software for the install media understands the requirements of the limited size of install media. For example, that could be mean different compile time options or creating subpackages to make it easier for tools like lorax to compose install images.

For compression, I think looking at zstd over xz is a good idea for the project.

This was discussed in today's meeting:
REJECTED (0, 0, -8)

Announced in the meeting summary.

I found that this proposal got rejected, because of concerns over the increased compression ratio.

But can we discuss accepting the change only partially -- that is removing the intermediate layer that is present on the SquashFS filesystem inside the DVD? This will reduce the image size by 24MiB without changing the compression options. Moreover, the SquashFS image can be made reproducible because of this change.

See the pull request for details: https://pagure.io/pungi-fedora/pull-request/888

Yes, I think eliminating ext4 layer is good. I think I even said that already in one of the mails in the fedora-devel thread. Please resubmit the change with just that part and you'll have my vote.

Hello Zbyszek,

I think I even said that already in one of the mails in the fedora-devel thread
That is right. I've seen your email message regarding this.

I submitted a new proposal: https://fedoraproject.org/w/index.php?title=Changes/OptimizeSquashFSOnDVDByRemovingEXT4FilesystemImageLayer

The change proposal covers only removing the EXT4 filesystem layer.

Hello Zbyszek,

Yes, I think eliminating ext4 layer is good. I think I even said that already in one of the mails in the fedora-devel thread. Please resubmit the change with just that part and you'll have my vote.

A FESCo issue for that change is already available at the following location: https://pagure.io/fesco/issue/2489