On Mon, Feb 03, 2020 at 11:11:27AM -0500, David Cantrell wrote:
In today's FESCo meeting (03-Feb-2020), we discussed this change proposal.
Being the engineering steering committee, we all had our own ideas and
opinions about what the problem is and how best to approach it. After
discussion, we agreed there are different goals related to this change
proposal and we as a community should agree on the goals and their priorities
before moving on to the implementation details.
The goals identified:
1) Reduce the ISO image size.
2) Improve installation time.
3) Improve image composition time.
We want input from the community on what the main goal should be and
prioritize the rest. For example, is ISO reduction size more important than
improving installation time, for instance? If so, why?
I think it's a balance, we shouldn't try to only focus on one and ignore
the rest, but have a priority list and try and balance the others based
Assuming we are starting from what we have now, IMHO, we should put
install time first, reducing image size next and only last put compose
I think that most users doing installs will expect downloads to take
some time (although not too long anymore with increased broadband
speeds), so will likely start the download and do something else until
it's done/they get back to it. Install times however are much more
visible to people as they are sitting there doing the install. Compose
times are something we ideally would reduce, but the only people really
waiting on composes are testers and the like, it really is pretty
divorced from end users.
So, I propose:
reduce install time
reduce image size
reduce compose time
is the order we should put them in, and I think switching to zstd and
tweaking things we can improve install time from where it is now, reduce
image size and hopefully not increase compose time much.