Varnish Cache project autumn cleaning

Poul-Henning Kamp phk at phk.freebsd.dk
Tue Nov 3 13:23:31 CET 2015


Hi everybody, a quick status update from the Varnish Cache project.


We got the 4.1 release out (finally!) and it doesn't look horrible
so you should start to plan your migration now.


HTTP is changing
----------------

The HTTP world is changing rapidly under us, HTTP2 wasn't the
cure-all some people thought it would be, so it is very likely
that we will see either a HTTP/3 or some significant improvements
to HTTP/2 and new transport protocols in the coming years.

The strategic plan for Varnish, to the extent we have one, is
to continue to be the best swiss-army-knife of HTTP, that
will not change.  But for reasons of man-power and general
sanity, we will not chase after every single Internet Draft
that has HTTP in the title.

Our next big goal is to implement the good parts of HTTP/2.


Future Releases
---------------

4.1-R has been a long, too long, time coming, for reasons which
I think any Open Source Project will experience at some point.

Fundamentally what has happened is that the devops who used to hack
Varnish have become full-time developers.

That is wonderful in 17 different ways, but it has one significant
down-side:  Getting things tested is much harder for us now.

It used to be that "somebody" would give -trunk or the release
candidate a spin on "their site" and get us some immediate 
feedback, but now we have to persuade somebody else to do this
and it adds a lot of impedance to the process.

So we fell into the "well, it's not tested enough, and while we
wait I can just add this..." loop and the deadlines whooshed by.

We're not making that mistake again.

In the future we are going to make two releases from the head of
the source tree every year, one in spring one in autumn and, with
the exception of the spring '16 release, you will know the next
release date at least six months in advance.

Whatever is in the tree, whatever works on the release date, is
what the release will contain.

This makes life much more predictable for everybody, developers
know when to be ready and tested, users can plan when to
decide on upgrades.

Not all of these releases will be equally useful, and only
some of them will be supported for the longer term.  That
decision will still have to be made case by case.


Tools
-----

We are also taking the opportunity to do some house-cleaning in the
project, to try to make the tools we use match the things we do
better.

We are going to drop the "Trac" tool and move to github, Lasse is
already busy figuring out the least painful way to do this.

We will also drop "patchwork", it's never really worked for us,
we'll try if github pull requests work, otherwise we'll find
something else.

The project homepage will also get a makeover soon-ish, with the
goal of making it more lively and relevant for the project and
users.

We're still pondering testing-frameworks (or as it's called these
days: Continuous Integration), but no decision has been made yet.

The mailing lists will stay the same.


Growing up, moving out
----------------------

Linpro, later Redpill-Linpro and now Varnish-Software gave birth
and have been doting and protective parents to the Varnish-Cache
FOSS project, but the time has come to start moving the project
away from home.

Let me make it absolutely clear that no amount of thanks can ever
be enough, and that there are no hard feelings involved.

The project is simply growing up, and needs to grow up more, and
it needs to develop its own identity and life.

Such a process is entirely natural and healthy, and like many young
people have, the Varnish-Cache Project will discover that the reason
the gutters didn't use to run over, was that Daddy was up there on
a ladder twice a year making sure they didn't.

Daddy will still be around, Varnish-Software is still entirely
committed to the Varnish-Cache FOSS project, but there will be
more stuff to do for ourselves and hopefully more people will
join in and help us do them.

In particular we *really* need somebody to work in the documentation,
and there are only so many hours (28 or something, right?) in a day.


Meetings
--------

There will be a Varnish meeting in Rotterdam on december 3rd/4th.
First day for users, second day a hackathon for devs/hackers/contributors,
put a mark in your calendar -  details will emerge shortly.

Hope to see you there...


And finally
-----------

I want to conclude with a big Thank You! to the Varnish Moral
License holders, which fund my Varnish habit:  It wouldn't happen
without your support.


Poul-Henning

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.



More information about the varnish-announce mailing list