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: Release candidate 2: curl 8.13.0-rc2
- Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ] [ by messages with attachments ]
From: Samuel Henrique via curl-library <curl-library_at_lists.haxx.se>
Date: Wed, 26 Mar 2025 21:59:24 +0000
Hello Jeroen, I'm cc'ing the Debian Science team here, as they maintain
r-cran-curl.
On Sun, 23 Mar 2025 at 09:45, Jeroen Ooms <jeroenooms_at_gmail.com> wrote:
>
> A Debian user has notified me that this update causes a unit test to
> fail in the R bindings. I looked into it and this is indeed the case,
> though it is harmless. The test was checking verbose messages for the
> presence of "existing connection" when re-using a handle, and
> seemingly the new version of libcurl changed this wording into
> "Re-using existing https: connection". So I'll relax that text a bit.
>
> It would be interesting though to see understand Debian CI did not
> reveal this. I think they only run a subset of the tests.
The Debian tests are being done calling:
LC_ALL=C.UTF-8 R --no-save < testthat.R
https://salsa.debian.org/r-pkg-team/r-cran-curl/-/blob/f2a331e666a6e59d38598b14992e42e758898040/debian/tests/run-unit-test#L17
The logs from autopkgtest show the following:
29s autopkgtest [18:57:38]: test run-unit-test: [-----------------------
29s
29s R version 4.4.3 (2025-02-28) -- "Trophy Case"
29s Copyright (C) 2025 The R Foundation for Statistical Computing
29s Platform: x86_64-pc-linux-gnu
29s
29s R is free software and comes with ABSOLUTELY NO WARRANTY.
29s You are welcome to redistribute it under certain conditions.
29s Type 'license()' or 'licence()' for distribution details.
29s
29s R is a collaborative project with many contributors.
29s Type 'contributors()' for more information and
29s 'citation()' on how to cite R or R packages in publications.
29s
29s Type 'demo()' for some demos, 'help()' for on-line help, or
29s 'help.start()' for an HTML browser interface to help.
29s Type 'q()' to quit R.
29s
29s > library(testthat)
30s > library(curl)
30s Using libcurl 8.13.0-rc1 with OpenSSL/3.4.1
30s >
30s > # Comply with CRAN policy: skip tests if no internet
30s > if(curl::has_internet()){
30s + test_check("curl")
30s + }
30s This is libcurl version 8.13.0-rc1 with OpenSSL/3.4.1
30s Using test server: https://hb.r-universe.dev
40s [ FAIL 0 | WARN 0 | SKIP 4 | PASS 146 ]
40s
40s ══ Skipped tests (4) ═══════════════════════════════════════════════════════════
40s • IDN tests fail on autopkgtest infrastructure (1): 'test-idn.R:3:3'
40s • On CRAN (3): 'test-multi-download.R:14:3', 'test-seek.R:2:3',
40s 'test-winssl.R:2:3'
40s
40s [ FAIL 0 | WARN 0 | SKIP 4 | PASS 146 ]
40s >
Linking to a previous execution, but all of them succeed:
https://ci.debian.net/packages/r/r-cran-curl/unstable/amd64/58601562/
Based on a recent commit from you, I assume this is this test that should have
failed:
https://github.com/jeroen/curl/commit/b1f1a109f30dc802df4394b7679d59f3092a0835
And the problem now is that the unit tests don't provide enough information for
us to confirm whether that test ran or not, it just says "PASS 146".
It's likely that that specific test is not being run for some reason, so how do
we know which tests are being run?
Is this an issue with calling the tests the wrong way, or is it something that
needs to be improved upstream?
Cheers,
Date: Wed, 26 Mar 2025 21:59:24 +0000
Hello Jeroen, I'm cc'ing the Debian Science team here, as they maintain
r-cran-curl.
On Sun, 23 Mar 2025 at 09:45, Jeroen Ooms <jeroenooms_at_gmail.com> wrote:
>
> A Debian user has notified me that this update causes a unit test to
> fail in the R bindings. I looked into it and this is indeed the case,
> though it is harmless. The test was checking verbose messages for the
> presence of "existing connection" when re-using a handle, and
> seemingly the new version of libcurl changed this wording into
> "Re-using existing https: connection". So I'll relax that text a bit.
>
> It would be interesting though to see understand Debian CI did not
> reveal this. I think they only run a subset of the tests.
The Debian tests are being done calling:
LC_ALL=C.UTF-8 R --no-save < testthat.R
https://salsa.debian.org/r-pkg-team/r-cran-curl/-/blob/f2a331e666a6e59d38598b14992e42e758898040/debian/tests/run-unit-test#L17
The logs from autopkgtest show the following:
29s autopkgtest [18:57:38]: test run-unit-test: [-----------------------
29s
29s R version 4.4.3 (2025-02-28) -- "Trophy Case"
29s Copyright (C) 2025 The R Foundation for Statistical Computing
29s Platform: x86_64-pc-linux-gnu
29s
29s R is free software and comes with ABSOLUTELY NO WARRANTY.
29s You are welcome to redistribute it under certain conditions.
29s Type 'license()' or 'licence()' for distribution details.
29s
29s R is a collaborative project with many contributors.
29s Type 'contributors()' for more information and
29s 'citation()' on how to cite R or R packages in publications.
29s
29s Type 'demo()' for some demos, 'help()' for on-line help, or
29s 'help.start()' for an HTML browser interface to help.
29s Type 'q()' to quit R.
29s
29s > library(testthat)
30s > library(curl)
30s Using libcurl 8.13.0-rc1 with OpenSSL/3.4.1
30s >
30s > # Comply with CRAN policy: skip tests if no internet
30s > if(curl::has_internet()){
30s + test_check("curl")
30s + }
30s This is libcurl version 8.13.0-rc1 with OpenSSL/3.4.1
30s Using test server: https://hb.r-universe.dev
40s [ FAIL 0 | WARN 0 | SKIP 4 | PASS 146 ]
40s
40s ══ Skipped tests (4) ═══════════════════════════════════════════════════════════
40s • IDN tests fail on autopkgtest infrastructure (1): 'test-idn.R:3:3'
40s • On CRAN (3): 'test-multi-download.R:14:3', 'test-seek.R:2:3',
40s 'test-winssl.R:2:3'
40s
40s [ FAIL 0 | WARN 0 | SKIP 4 | PASS 146 ]
40s >
Linking to a previous execution, but all of them succeed:
https://ci.debian.net/packages/r/r-cran-curl/unstable/amd64/58601562/
Based on a recent commit from you, I assume this is this test that should have
failed:
https://github.com/jeroen/curl/commit/b1f1a109f30dc802df4394b7679d59f3092a0835
And the problem now is that the unit tests don't provide enough information for
us to confirm whether that test ran or not, it just says "PASS 146".
It's likely that that specific test is not being run for some reason, so how do
we know which tests are being run?
Is this an issue with calling the tests the wrong way, or is it something that
needs to be improved upstream?
Cheers,
-- Samuel Henrique <samueloph> -- Unsubscribe: https://lists.haxx.se/mailman/listinfo/curl-library Etiquette: https://curl.se/mail/etiquette.htmlReceived on 2025-03-26