Dear fellow people,

I worked quite a bit on Devuan stuff the past couple days, specifically
files.devuan.org mirrors.

Here are the results:


Update to devuan-editors/devuan-www/data/mirrors.yml
====================================================
I went through IRC logs of #devuan and identified some mirrors that are
not listed anywhere. Did you know the LMU in Munich has had a mirror for
about a year now? Or that Princeton's mirror supports RSYNC?

Should I add this mirror?
http://devuan-temporary-mirror.almesberger.net/files.devuan.org/

All of this and a bit more is reflected here:
https://git.devuan.org/evilham/devuan-www/blob/beta.devuan.org/data/mirrors.yml

I reordered the entries so that they reflect a somewhat chronological
order which, according to golinux, is what is being reflected on
devuan.org right now.

Removed unnecessary path / status entries, standardised bandwidth units
so we can sort by that field if necessary, added mirrors that were
either nor listed here but in devuan.org or not listed anywhere, added
protocols that are supported but not documented, added the issue /
maintainer where possible/needed.

This brings us to...


Mirror state page
=================
You may remember I set up a status page a few weeks ago for fdo mirrors,
it is currently at: http://devuan.evilham.org.

It now supports all protocols and updates every 2 hours. It has been
vastly improved visually and gives us a good overview of the mirrors.

The source for the mirror data is the aforementioned mirrors.yml file; I
pull my personal fork, check the mirrors and generate the static
index.html and the status.json files that are served there.

While doing that, I also change my source/pages/index.en.html.md.erb to
reflect any changes in the mirror, commit that, and push the changes
back to my personal fork.

This means: changing the mirrors.yml file could be used to trigger the
update of the mirrors list, allowing for change to take place in *one
place* and by separating data from its representation, for traceability
+ documentation.

Again, did we know about the LMU mirror? If not, it's kind of sad that
they have been offering it for a year and from what I've read, it would
have been very useful to a mirror like this have back then. But water
under the bridge.
https://botbot.me/freenode/devuan/msg/71016722/

A diff of the index.en.html.md.erb file on my fork and that on the site
shows that the changes are, besides new entries, a standardisation of
trailing slashes.

On that note...


New mirrors
===========
Two mirrors were added these days, one of them from the Aalborg
University in Denmark. https://mirrors.dotsrc.org/devuan/
The other one comes from what seems to be a helpful devuanite (onefang).
https://sledjhamr.org/devuan/

Golinux has updated the devuan-editors/devuan-www repo, the page should
be regenerated.


Past Mirror offerings
=====================
While checking the logs, I came across a few mirror offerings, this one
in Canada seemed particularly serious business (University of Waterloo):
https://botbot.me/freenode/devuan/2015-08-04/?msg=46419869&page=3

Maybe we should drop them an email. Amongst other things, they have a
full debian mirror.


GitLab Issue closing
====================
There were a couple issues relating an inactive selea.se mirror; they
expressed interest in hosting one in the future, but right now the
mirror does not exist. So those issues were closed by golinux already.

Issue #149 relating a new mirror can probably be closed already
https://git.devuan.org/devuan-editors/devuan-www/issues/149#note_9829


Next?
=====
When it comes to documenting fdo mirrors, only some information is
missing and the IRC logs (or maybe its search) didn't help me there; so
there's nothing I can do.

The mirrors state site is open for constructive comments, maybe we can
integrate it somewhere at some point since it provides more information
for users (e.g. server location, bandwidth and if it's available).
I specifically did not use anything that was not being used on the main
website already, so that shouldn't be a problem.

The integration of my script and the devuan-www repo is optional, but I
would think it makes sense to have the information come from a file
whose structure is not going to change; changing the way we represent
mirrors become a breeze and we also have a good authoritative source for
that information.

That being said, since the other general project issue I had spotted is
being worked on by someone else (according to 2 weeks ago pad; it was
about using amprolla output to list package status); I'm exhausting this
topic and would start moving on to something else.

That something else could be somewhat related (fdo mirrors? haven't seen
much information on that) or maybe packaging. I think packaging is one
of the things Devuan is needing the most right now, so if someone is
around next week after 13:00 UTC and could help me when I get stuck, it
would be great.
Remember I have a tendency to document stuff, so it would probably help
improve current documentation and, in the long run, help with getting
new packages maintainers.


Looking forward to your comments,
--
Evilham