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.
Please consider a bugfix release (8.23.1)
- Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ] [ by messages with attachments ]
From: Alexander Turenko via curl-library <curl-library_at_lists.haxx.se>
Date: Fri, 23 Oct 2026 03:06:49 +0300
Hi!
I'd like to highlight two unreleased bugfixes for the libcurl
maintainers to decide if they are worth a bugfix release.
The 8.22.0 release introduced a regression where bytes sent by the
server and pending in a socket may be taken as a response to the next
request. The regression was introduced in
https://github.com/curl/curl/pull/22169 and fixed in
https://github.com/curl/curl/pull/23412. The fix hasn't landed in any
release yet.
Honestly, I can't judge its impact. At first it looked like a timing
issue anyone may hit (and, what is worse, it reproduces intermittently).
Now, I read it as a case of incorrect server behavior.
The second problem is the one we (in the tarantool team) hit ourselves.
I reported it in https://github.com/curl/curl/issues/23445: an early
response and close from the server leads to CURLE_COULDNT_CONNECT on
macOS (the same transfer succeeds on Linux). It is fixed now and the fix
is not released yet.
It is also a timing issue. The server behaves correctly (I guess), but
the scenario is not very usual (except, likely, in tests, where we hit
it).
I'm asking you to consider if these (and the other bugfixes on master)
are worth a bugfix release before the feature release planned for
December.
Personally, I'm fine with staying on a commit from master (but of course
would prefer an official release). However, others may gain from the
release too: at least Homebrew's users would get a macOS-specific fix.
WBR, Alexander Turenko
Date: Fri, 23 Oct 2026 03:06:49 +0300
Hi!
I'd like to highlight two unreleased bugfixes for the libcurl
maintainers to decide if they are worth a bugfix release.
The 8.22.0 release introduced a regression where bytes sent by the
server and pending in a socket may be taken as a response to the next
request. The regression was introduced in
https://github.com/curl/curl/pull/22169 and fixed in
https://github.com/curl/curl/pull/23412. The fix hasn't landed in any
release yet.
Honestly, I can't judge its impact. At first it looked like a timing
issue anyone may hit (and, what is worse, it reproduces intermittently).
Now, I read it as a case of incorrect server behavior.
The second problem is the one we (in the tarantool team) hit ourselves.
I reported it in https://github.com/curl/curl/issues/23445: an early
response and close from the server leads to CURLE_COULDNT_CONNECT on
macOS (the same transfer succeeds on Linux). It is fixed now and the fix
is not released yet.
It is also a timing issue. The server behaves correctly (I guess), but
the scenario is not very usual (except, likely, in tests, where we hit
it).
I'm asking you to consider if these (and the other bugfixes on master)
are worth a bugfix release before the feature release planned for
December.
Personally, I'm fine with staying on a commit from master (but of course
would prefer an official release). However, others may gain from the
release too: at least Homebrew's users would get a macOS-specific fix.
WBR, Alexander Turenko
-- Unsubscribe: https://lists.haxx.se/mailman/listinfo/curl-library Etiquette: https://curl.se/mail/etiquette.htmlReceived on 2026-10-23