Buy commercial curl support. We
help you work out your issues, debug your libcurl applications, use the API,
port to new platforms, add new features and more. With a team lead by the
curl founder Daniel himself.
Re: First baby steps at performance testing
- Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ] [ by messages with attachments ]
From: Daniel Stenberg via curl-library <curl-library_at_lists.haxx.se>
Date: Wed, 12 Aug 2026 09:03:33 +0200 (CEST)
On Tue, 11 Aug 2026, Daniel Stenberg via curl-library wrote:
> 4. The output is magically transferred to the curl website on a regular
> interval and appears on https://curl.se/perf/
Now all builds done using the same commit hash are "grouped" in the graph. The
main value (used for the main blue plot line) is the median value for all runs
on that same commit, and then there is a minimum and a maximum line for those
corresponding values. I'll continue to tweak the look to try out different
things.
The graph gets a little busier this way, but this now allows us to add many
more builds/commits into a single graph. There is going to be a maximum number
of data points before it gets too crowded. I suspect the biggest drawback with
this approach is if we update a dependency or something that changes
performance. But that's not going to be too frequent and risks disturbing
graphs no matter what.
I also introduced an "alias" system for the commits, so that they can be
refered to as C[number] on this page to make them readable in the graphs - and
moved the labels to the X axis. I now also provide GitHub links for those
commits at the top.
The moving average is now based on the main value from the 4 most recent
commits. As we get more commits into the graph I'll try different numbers.
Date: Wed, 12 Aug 2026 09:03:33 +0200 (CEST)
On Tue, 11 Aug 2026, Daniel Stenberg via curl-library wrote:
> 4. The output is magically transferred to the curl website on a regular
> interval and appears on https://curl.se/perf/
Now all builds done using the same commit hash are "grouped" in the graph. The
main value (used for the main blue plot line) is the median value for all runs
on that same commit, and then there is a minimum and a maximum line for those
corresponding values. I'll continue to tweak the look to try out different
things.
The graph gets a little busier this way, but this now allows us to add many
more builds/commits into a single graph. There is going to be a maximum number
of data points before it gets too crowded. I suspect the biggest drawback with
this approach is if we update a dependency or something that changes
performance. But that's not going to be too frequent and risks disturbing
graphs no matter what.
I also introduced an "alias" system for the commits, so that they can be
refered to as C[number] on this page to make them readable in the graphs - and
moved the labels to the X axis. I now also provide GitHub links for those
commits at the top.
The moving average is now based on the main value from the 4 most recent
commits. As we get more commits into the graph I'll try different numbers.
-- / daniel.haxx.se || https://rock-solid.curl.dev -- Unsubscribe: https://lists.haxx.se/mailman/listinfo/curl-library Etiquette: https://curl.se/mail/etiquette.htmlReceived on 2026-08-12