[Daniel's week] September 13, 2024

[Daniel Stenberg]

Hello.

Another completely busy work week ends.

## graphs

I started this week adding the author activity graph [1] to the curl
dashboard. It shows how many authors that have authored commits within a few
different sliding time frames: 120 days, 60 days, 30 days and 7 days. The
general appearance is roughly to one of those heart activity things you
sometimes see in hospital TV series. Or maybe polygraph results.

Conclusions are hard to draw from that graph I believe, other than perhaps
that we have rather steady amount of different authors involved.

I also added graphs that displays commit distribution on day of the year [2],
month of the year [3] and day of the week [4]. Turns out we make curl on
Mondays in August. At 74 separate graphs.

## curl

The beginning of the week was all preparations for the Wednesday curl release
and it was indeed one of those releases where we had things handled and in
order in proper time so the release felt relaxed and controlled. Even if we of
course knew there was a potential risk that there could be regressions, simply
based on the huge amount of changes and bugfixes we had landed this time.

I shipped the release [7] , posted the security advisory [8] and I managed to
do the release presentation video [9] live-stream before the first regressions
were reported. I had a short moment of release bliss before the harsh reality
hit me in the face.

Argh. It soon became painfully obvious that there were more than just one
annoying regression: we had shipped a whole set. It of course also worked to
remind ourselves of the eternal truth: you never get the real bugreports until
you ship a release. We think we have test cases, but no matter how many we
have there are always users finding different angles and twists to do things
that we have no test cases for. Yet.

We quickly decided to do a follow-up patch release to shorten the time window
users have to endure these regressions. We give it seven days so that we can
collect and merge as many bugfixes as possible and at the same time not rush
anything, and do a 8.10.1 release on September 18 next week. We have merged at
least thirteen bugfixes already.

## trurl

In a little breathing pause in the curl regression fixing, I headed over and
shipped a fresh trurl release as we had built up a few bugfixes in there
recently. I decided to call it 0.15.1 [6] as the first patch release made for
trurl.

Here as well a new release triggered activity and we soon had some new bugs
reported - that soon were fixed. Some of those reports triggered me to add
more tests, which led me to find a few more bugs that I fixed that made me
decide I should document some of the URL details better. How the different URL
components work with curl. While working on that I discovered a few more bugs
that I also fixed. It also resulted in a larger, improved, trurl manpage [10].

Clearly time to do yet another trurl release. I think I will give that seven
days for additional fixes to accumulate as well.

For the next trurl release I introduce a --qtrim option [15] that replaces the
former one called --trim. The new one is simply decided to only trim query
parts as the former one was designed to be able to trim other components as
well but we could never think of any. It just made it a little inconvenient to
use. --qtrim fixes that.

## DDos

I published yet another weirdo email [11] that I received the other day and I
found now when I was scanning through my spam folder.

## Sponsors

The curl sponsors page [12] was updated. We added two prominent top sponsors
that help funding the curl project: GitHub and IBB. I also tweaked the page
somewhat to look a little better on narrow displays like mobile phones.

## QUIC in the kernel

Back in July, the PR [13] for adding support for QUIC in the kernel to curl
was created, and there has been progress made in there ever since.

This week, that team submitted their kernel patch for QUIC to the netdev
mailing list [14].

## Coming up

- Wednesday curl 8.10.1 release
- Thursday trurl 0.16 release
- Friday meeting with my Polhem Prize council fellows

## Links

[1] = https://curl.se/dashboard1.html#authors-active
[2] = https://curl.se/dashboard1.html#daily-commits
[3] = https://curl.se/dashboard1.html#monthly-commits
[4] = https://curl.se/dashboard1.html#weekly-commits
[6] = https://daniel.haxx.se/blog/2024/09/12/trurl-0-15-1/
[7] = https://daniel.haxx.se/blog/2024/09/11/curl-8-10-0/
[8] = https://curl.se/docs/CVE-2024-8096.html
[9] = https://youtu.be/jO5qZ9VW0_k
[10] = https://curl.se/trurl/manual.html
[11] = https://bagder.github.io/emails/2024/2024-09-07.html
[12] = https://curl.se/sponsors.html
[13] = https://github.com/curl/curl/pull/14313
[14] = https://lists.openwall.net/netdev/2024/09/10/31
[15] = https://github.com/curl/trurl/pull/364

-- 

  / daniel.haxx.se