https://issues.dlang.org/show_bug.cgi?id=21130

Walter Bright <bugzilla@digitalmars.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |TestSuite

--

https://issues.dlang.org/show_bug.cgi?id=21130

Mathias LANG <pro.mathias.lang@gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
                 CC|                            |pro.mathias.lang@gmail.com
         Resolution|---                         |INVALID

--- Comment #1 from Mathias LANG <pro.mathias.lang@gmail.com> ---
Please don't open bugs for testsuite transient failure without identifying
them.
What can one do here to fix this bug ? Absolutely nothing. There's no
information, so this will either get ignored or closed, but never fixed.

For reference, we know that the Github actions job sometimes have some random failures, but they are trivial to re-trigger, and it's rather rare. We took steps to reduce it (by introducing caching, and hence reducing network load) but there's only so much we can do (some failures are on Github side).

--

https://issues.dlang.org/show_bug.cgi?id=21130

Walter Bright <bugzilla@digitalmars.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|INVALID                     |---

--- Comment #2 from Walter Bright <bugzilla@digitalmars.com> ---
(In reply to Mathias LANG from comment #1)
> [...]

There being "no information" is part of the bug report. That's all the log says. The logger can do better.

Secondly, these networking errors can be detected, and when they happen, the test script can "sleep" for a minute or two and then retry it. If it continues to fail after N times, then the script can give up and quit.

After all, if the user can re-start the script, so can the script. The script ought to be able to distinguish between an environmental error and a failure of whatever is being tested.

--

https://issues.dlang.org/show_bug.cgi?id=21130

moonlightsentinel@disroot.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
                 CC|                            |moonlightsentinel@disroot.o
                   |                            |rg
         Resolution|---                         |WONTFIX

--- Comment #3 from moonlightsentinel@disroot.org ---
There is nothing we can do here due to the missing log (which seems to be an issue inside of GitHub Actions).

Note that https://github.com/dlang/dmd/pull/11537 dropped the support for older llvms which should improve caching and reduce the chance of network error (which predominantly happened for those older versions).

--