[Daniel's week] June 7, 2024

[Daniel Stenberg]

Hi friends!

Here is a brief summary of my week:

## survey

I have worked on the curl user survey 2024 analysis. It is a lot of work. I
try to actually look hard at the answers, numbers and the resulting graph to
see if I can figure something out from them and then try to write that down.
As I only spend a few hours on this work a little on and off as it is quite
draining, it might just take me another week or two to complete.

## features

This week we opened the curl feature window and we have certainly merged a
good set already this week. The counter now says 7 changes and 92 bugfixes
committed in the 16 days since 8.8.0 shipped. Among the changes are two new
curl command line options: --ip-tos [5] for setting the IPv4 Type of Service
IP header value or the IPv6 Traffic Class one, and --mptcp [6] for enabling
Multipath TCP when curl runs on Linux.

As we did all this only on the first week out of three of the feature window,
I suspect this release cycle can end up quite a feature packed one.

## website language

While browsing an old CVE description I noticed language that we otherwise
don't use in the project. This triggered me to add another CI job to the web
site's GitHub repo to scan for "bad words" in the same way we do for the
source repository. It warns for use of contractions and words we generally try
to avoid or stick to. Like how we write "hostname" instead of "host name" and
a whole range like that. The idea being to use a more consistent English.

Made me correct more than 200 little language nits on the website. Mostly in
old vulnerability advisories.

## c-ares

On the topic of websites, this week I updated the c-ares DNS data to instead
make the c-ares entries point to a set of GitHub IP addresses as the hosting
of the website transitioned away from my server [1]. I have been hosting the
site for about twenty years by now so I figure it is about time someone else
gets to do it.

I still host the c-ares mailing list and I run the DNS server so I'm not
totally off the hook, but that's all fine. I don't need to detach myself
completely. I still care about the project and I still plan to follow it
closely, even if I have not participated actively much since a while back.

## webinar

On Thursday next week, June 13, I will do a webinar called "advanced libcurl"
[2]. It is going to be about using libcurl for internet transfers after you've
gotten a hang of the basics. Something like a step two class after the libcurl
for beginners session. With a Q&A session at the end to get all your questions
answered. Free of charge, just sign up.

## graph

The new graph of the week shows the number of authors left (show by git blame)
in production code per 1,000 lines of code (KLOC) [3]. I call it "author
density", and it shows that we have been above 3.5 authors/KLOC in curl for
the last couple of years.

## cookie length

I proposed that the upcoming updated cookie spec (rfc6265bis) should document
something about the maximum header length a HTTP client should send and the
minimum length a server should accept [4].

## Coming up

- continue work on the survey analysis
- Thursday: deliver the libcurl webinar

## Links

[1] = https://daniel.haxx.se/blog/2024/06/06/bye-bye-hosting-c-ares-web/
[2] = https://us02web.zoom.us/webinar/register/9316747720453/WN_zudyMou_RviqdgL-3QQXeg#/registration
[3] = https://curl.se/dashboard1.html#density
[4] = https://github.com/httpwg/http-extensions/issues/2797
[5] = https://curl.se/docs/manpage.html#--ip-tos
[6] = https://curl.se/docs/manpage.html#--mptcp

-- 

  / daniel.haxx.se