[Pkg-rust-maintainers] Rust Packages, BoF meeting notes

[Ximin Luo]

Sean Whitton:
> Hello Ximin,
> 
> On Mon 19 Oct 2020 at 08:34pm +01, Ximin Luo wrote:
> 
>> I seem to remember that we did some calculations and the sizes are not
>> significantly affected by these empty packages. For every empty rust
>> package, there is >1 non-empty rust package. A 50% metadata bloat of
>> what is probably ~2-3% of the whole Debian archive is not a big deal,
>> and bandwidth is constantly getting cheaper. 5G and/or fiber is coming
>> to a postcode near you within the next few years I am sure.
> 
> We have lots of users in various parts of the world with much worse
> Internet connections than the average DD.  It would be good to continue
> to support those users where we can.
> 

I believe my proposal supports those users just fine. Let's crunch some numbers.

1. Automatically passing these empty rust packages (as I suggest) might slow down every Debian user perhaps by O(1 second).
2. Holding these empty rust packages slows down every user of Debian rust packages perhaps by O(several weeks).
3. Blocking these empty rust packages completely, slows down every user of Debian rust packages perhaps by O(several months) or even longer.

There are 604800 seconds in a week.

The current situation wastes more time for users overall, unless you believe that:

- {Debian users} outnumber {users of Debian rust packages} by a factor of 604800 to 1, or
- users of Debian rust packages are much more likely to have faster internet connections, and this difference in probabilities outweighs Debian users by a factor of 604800 to 1.

neither of which seem like reasonable beliefs to me.

Ximin

-- 
GPG: ed25519/56034877E1F87C35
https://github.com/infinity0/pubkeys.git