[Pkg-cmake-team] Bug#1126187: cmake: FTBFS on i386: test 535 - RunCMake.string (Failed)

Drew Parsons dparsons at debian.org
Fri Jan 23 09:27:31 GMT 2026


Package: cmake
Version: 4.2.2-1
Severity: serious
Tags: ftbfs
Justification: FTBFS
User: debian-qa at lists.debian.org
Usertags: i386

cmake 4.2.2 is failing buildtime tests on i386

526/735 Test #535: RunCMake.string .............................................***Failed    1.11 sec
-- JSON - PASSED
-- JSONNoJson - PASSED
-- JSONWrongMode - PASSED
-- JSONOneArg - PASSED
-- JSONNoArgs - PASSED
-- JSONBadJson - PASSED
-- Append - PASSED
-- AppendNoArgs - PASSED
-- Prepend - PASSED
-- PrependNoArgs - PASSED
-- Concat - PASSED
-- ConcatNoArgs - PASSED
-- Join - PASSED
-- JoinNoArgs - PASSED
-- JoinNoVar - PASSED
-- Timestamp - PASSED
CMake Error at /build/reproducible-path/cmake-4.2.2/Tests/RunCMake/RunCMake.cmake:297 (message):
  Timestamp2038 - FAILED:

  Result is [27], not [0].

  stderr does not match that expected.

  Command was:

   command> "/build/reproducible-path/cmake-4.2.2/Build/bin/cmake" "/build/reproducible-path/cmake-4.2.2/Tests/RunCMake/string" "-G" "Unix Makefiles" "-DRunCMake_TEST=Timestamp2038" "--no-warn-unused-cli" "-DCMAKE_MAKE_PROGRAM=/usr/bin/gmake"

  Actual stdout:

   actual-stdout> Not searching for unused variables given on the command line.

  Expected stderr to match:

   expect-stderr> ^RESULT=2038-01-20 05:00:00\.000000 Wednesday=Wed January=Jan 38 day=020 wd=3 week=03 w_iso=03 %I=05 epoch=2147576400 TZ=GMT tz=\+0000$

  Actual stderr:

   actual-stderr> CMake Error: Cannot parse SOURCE_DATE_EPOCH as integer

Call Stack (most recent call first):
  /build/reproducible-path/cmake-4.2.2/Tests/RunCMake/string/RunCMakeTest.cmake:29 (run_cmake)


The following tests FAILED:
	535 - RunCMake.string (Failed)
Errors while running CTest



It looks like a t64 (2038) error.  It's testing for timestamp 2038
which is out of bounds on a 32-bit time count. In that case perhaps
the Timestamp2038 test should be skipped on 32-bit systems.

On the other hand the RunCMake.string test is passing on other 32-bit
systems, armhf, powerpc. What is i386 doing differently with respect
to this test compared to the other 32-bit architectures?



More information about the Pkg-cmake-team mailing list