This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[Fwd: Commit approval pls: PR 9877, 8336, plus other stuff]


Hi Kean,

For fixinc stuff, please CC me or put "fixinc" in the subject.
I don't read everything, but I chanced upon this.

>  * fixinc/check.tpl: Allow user to specify diff program for make check
OK.

>  * fixinc/inclhack.def: Fix several SCO header files, namely string.h,
>    math.h, sys/byteorder.h and sys/regset.h.
AAB_svr4_replace_byteorder: you are *introducing* platform constraints,
not adding ``mach = "i[34567]86-*-sco3.2v5*";'' and I do not see a purpose
to the cosmetic change to htonl.

sco_math:  Are you sure the test_text triggers?  If it does, then the "select"
clause is too generic because the select clause will not match the test_text.
Are you sure this fix will only fire on SCO-problematic headers?  ``inline double abs''
seems like a weak test on math headers.  Continuing with test_text: you do have
two fixes here.  One for __fp_class and the other for disabling inline functions
``abs'' through ``trunc''.  The test should cover both issues, too.

sco_regset:  please remove several of the ``\n''-s.  If a ripple shows up in
the result, I'd like the ``diff -c'' to cover the test name, too.  (i.e., the
``#ifndef SCO_REGSET_CHECK'' and ``#endif /* SCO_REGSET_CHECK */'' are really
visual tags that ought to show up in a ``diff -c''.)

>  * fixinc/tests/base/math.h: Update

You'll notice that the fix was not applied.
Please examine the results of the "make check" and check that
what you intended to have happen has, in fact, happened.

>  * fixinc/tests/base/sys/byteorder.h: Update
Unnecessary.

>  * fixinc/tests/base/string.h: Added

Notice again, your fix did not fire.  This is an artifact of the testing scheme,
tough.  Your ``sco_string'' fix causes the file to be added to the tests, but
AAB_ultrix_string is a replacement fix that suppresses yours.  Fix this by
moving another of the headers to fix to the start of the list.  e.g.:

>  files = ansi/string.h, posix/string.h,  string.h,
>          xpg4/string.h, xpg4v2/string.h, xpg4plus/string.h,
>          ods_30_compat/string.h, oldstyle/string.h;

>  * fixinc/tests/base/sys/regset.h: Added
I'd be happier if the result were more like:

+#if defined( SCO_REGSET_CHECK )
+union u_fps {
+ struct rsfpstate
+ { int whatever; } };
+union _u_fps {
+ struct _rsfpstate
+ { int whatever; } };
+#endif /* SCO_REGSET_CHECK */

Don't sweat GNU coding styles for the testing text.  It's not real code. :-)

Thanks, Kean.  Regards, Bruce


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]