Prof A Olowofoyeku (The African Chief) a écrit:
On 3 May 2003 at 23:50, Jaroslaw J. Pyszny wrote:
[...]
My test (all ok only one warning):
Testing xgpc 20030323, based on gcc-3.2.3 (athlon-pc-linux) (lx.siec), [...] TEST fjf386.pas: gpc1: warnings being treated as errors fjf386.pas: In procedure `baz': fjf386.pas:11: warning: statement with no effect ^^^^^^^^^^^^^^^^^^^^^^^^^______
I got 2 failures under Mingw (I am running test_run directly). The failures look spurious to me:
TEST dostest.pas: gpc1.exe: warnings being treated as errors ./../units/system.pas: In function `MemAvail': ./../units/system.pas:634: warning: capitalisation of identifier `T' doesn't match ./../rts/gpc.pas:1620: warning: previous capitalisation `t' gpc1.exe: c:\mingw\bin\gpc.exe exited with status 1 failed
TEST fjf437.pas: gpc1.exe: warnings being treated as errors ./../units/system.pas: In function `MemAvail': ./../units/system.pas:634: warning: capitalisation of identifier `T' doesn't match ./../rts/gpc.pas:1620: warning: previous capitalisation `t' gpc1.exe: c:\mingw\bin\gpc.exe exited with status 1 failed
This is a consequence of the new warnings on different capitalisations. Run with -Wno-identifier-case (e.g. through EXTRA_TEST_PFLAGS) to eliminate. I have realized recently with horror than "identifier name conflicts" is really between identifier names, even if for pascal they are completely unrelated objects, i.e. two local identifiers in two different modules, like in this case. This is completely against any modularity concept. So I have replaced eveywhere in my system -Wall by -Wall -Wno-identifier-case When seeing the related discussions on the list, I had understood that the warning was for different capitalizations of the identifier of the same pascal object, which is useful, at least from the aesthetic point of view.
Maurice