2020 ideas
Ideas for talks and sessions for curl up 2020. Ideally with names next to them!
It would be very useful to get regular performance tests going to catch regressions or any general degradation.
How can we do this? What should we test? How? What other projects do this that we can draw experiences and/or tools and ideas from?
(Daniel)
If we would gather all logs from all tests from merged code, we could start analyzing which tests that fail and how often and thereby detect flaky tests using statistics.
We could also then start actually running (but ignoring the results of) flaky tests, so also detect if previously detected and marked flaky tests become more stable or otherwise change "pattern".
(Daniel)
With sponsor money coming in and getting spent, the primary arbiter of deciding this is Daniel. Maybe it would be better to get a team together so that we could discuss and decide on these matters with a few more brains and viewpoints involved?
It might also feel safer for sponsors and contributors to know that we're not at the whims of a singular individual.
Who would become members of such a team? How would we decide who? What tasks should the core-team deal with?
(Daniel)
We have several sponsors in several levels in the project and money is coming in. We're spending funds on the bug bounty, funding attendees to curl up and so on. What more can we do to keep sponsors happy (and attract new ones) and identify targets to spend funds on that improves and furthers the project?
(Daniel)
A report from the security-team on how the Bug Bounty has fared, what we've learned and anything else relevant around curl security.
(maybe Daniel)
Real world demonstration of large scale data loading using the magic of parallel libcurl and docker to load/retrieve data into a massive database.
(Jim Fuller)
Last year we spent some time crafting a simple build system for producing a single docker image for release on docker hub. While the usage of curl in 'docker land' is somewhat confusing (and the future of docker itself 'cloudy') there seems to be some uptake which I will report on. I will also spend some time on design, process and build of the current docker image and outline possible 'next steps'.
(Jim Fuller)
Where's the protocol, other clients, servers and everything right now? What happens next?
Where's curl, the implementation, the docs, the tests and everything right now? What's next?
(Daniel)
Where we are. The good and the bad. What's to fix and where do we want to go next?
(maybe Daniel)
(Jeroen) Some successes and challenges of writing bindings for the R programming language.
Topics could include:
- What is the curl R package: A Modern and Flexible Web Client for R.
- How to safely let users set CURLOPT_ handle options at runtime.
- Challenges of writing portable bindings that work on all common libcurl versions/configurations/systems.
- Problems with the transition to HTTP/2
- Discussion: should the curl project advise distribution maintainers (Debian, Fedora, Apple) which version to ship for LTS releases and which patches to backport.
(Stefan Eissing)
Topics - if I make it to the workshop - could include:
- How Apache httpd nowadays uses multi-curl
- What OCSP server side stapling in TLS really is
- How https: on the internet almost melted while making it work (maybe a bit exaggerated, maybe not!)
Taming the jungle that is curl --help. How do you help people find what they need out of 230 command-line options?
(Dan Fandrich)