In #curl we have improved our TLS early-data support recently. Mostly @icing actually.
Today Robin and Rafel at Akamai timely blogged about how early-data speeds things up.
https://www.akamai.com/blog/edge/lightning-fast-requests-with-early-data
In #curl we have improved our TLS early-data support recently. Mostly @icing actually.
Today Robin and Rafel at Akamai timely blogged about how early-data speeds things up.
https://www.akamai.com/blog/edge/lightning-fast-requests-with-early-data
Welcome Zhaoming Luo as #curl commit author 1348: https://github.com/curl/curl/pull/16523
Adding #curl release candidates
https://daniel.haxx.se/blog/2025/02/28/adding-curl-release-candidates/
I know it is often repeated, but #curl is not a one man factory:
@jayalane Code often incorporates a lot of experience that are hard to port over to another language. This is why even Rust language creators don't recommend RiiR and created good FFI mechanisms instead.
Rust is an easy choice for new projects or pieces of a modular project. But some people do push it. I've seen the #Curl developer being asked to rewrite it in Rust - without recognizing how impractical that is.
One of the challenges with supporting 28 protocols in #curl is that people expect me to know all these protocols. - I do, but to varying degrees.
I think #curl needs a few minor tweaks to support this really well...
Got wcurl into @fedora and am now maintaining it!
https://github.com/curl/wcurl/releases/tag/v2025.02.24
Update builds rolling out today!
If you use a curl built with GnuTLS (debian sid, for example), specifying --ciphers now does nothing.
But some people have need of that option to work. Here you can chime in on how it should be supported:
A year ago the third #curl security audit completed, that time covering our HTTP/3 implementation.
It revealed **no** major discoveries or security problems.
https://daniel.haxx.se/blog/2024/02/23/curl-http-3-security-audit/
@bagder Or on average a net about +17 SLOC per day.
Because everyone knows More Code Is Better, and We Must Count The Kay-Locks! It's the Kay Pee Eye!
(KLOC, KPI)
#curl predecessor httpget 0.2 from around 1996/1997 is 165 lines. Needless to say, it has multiple critical security vulnerabilities. How many can you spot?
If you build it on a modern system and want to try exploiting it in true 90s fashion, be sure to turn off address space layout randomisation (ASLR).
Sunday surprise!
A friend of mine found an old email from me dated January 17 1997
Attached in this mail was the #httpget 0.2 source code. Previously believed to be lost, now the oldest httpget code I have.
165 lines long. 110 lines code, 30 lines comments, 25 blank lines.
This morning, #curl was 174,854 lines of code, not counting blank lines but comments.
1248 times larger over 28 years.
#curl website traffic Feb 2025
Data without logs sure leaves us open for speculations.
https://daniel.haxx.se/blog/2025/02/22/curl-website-traffic-feb-2025/
The #curl feature window is now open. Go go go.
Sometimes people think #curl is a simple little HTTP tool, while in reality there's a whole internet transfer machine in there supporting 28 protocols.
Wait, what?
#curl can send email? Damn... @bagder I'm impressed (not that I didn't expected that), curl just keeps on giving.
curl --url "smtp://$SMTP_SERVER:$SMTP_PORT" \
--ssl-reqd \
--mail-from "$SMTP_USER" \
--mail-rcpt "$EMAIL" \
--upload-file "$LOG_FILE" \
--user "$SMTP_USER:$SMTP_PASSWORD" \
--insecure
How I create and maintain RELEASE-NOTES for #curl: live-streamed tomorrow.
It was mostly implied from my how I do releases video, so I figured I could show a little more from how the sausage is actually made.
Should be a fairly brief one. At 10:00 CET tomorrow Friday over at https://www.twitch.tv/curlhacker as usual.
As amusing as it is, the pillow quote is not correct though. #curl is my full-time gig since over six years now.
This is what I do: https://daniel.haxx.se/job.html