#15 Eliminate warnings and errors with gcc-4.3-20071109

[Tom Browder]

I just got gcc-4.3-20071109 snapshot built and started to use it and soon discovered that it finds more problems than 4.2.1, including errors with loki-lib. I copied my loki-lib directory (which is synched with sourceforge) to loki-lib_gcc4.3 where I tried to eliminate all the errors and warnings (some warnings I ignored per Andrei's embedded instructions and for others I had to turn off
certain warning flags). I have attached a patch from running:

diff -Naur -x.svn loki-lib loki-lib_gcc4.3

-Tom

Tom Browder
Niceville, FLorida
USA

[Zak Kipling]

Logged In: YES
user_id=764271
Originator: NO

Several of these patches add

#include <string.h>

in order to get global-namespace declarations of strlen, strcmp and friends.

Isn't this deprecated in Annex D of the standard, in favour of

#include <cstring>

with the names qualified as std::strlen etc?

[Tom Browder]

Logged In: YES
user_id=861162
Originator: YES

I believe you're correct. I'll go back, make those changes, make sure the changes compile, and attach the corrected patch.

[Tom Browder]

Logged In: YES
user_id=861162
Originator: YES

Hm, after updating my loki with svn, and making the changes from <string.h> to <cstring>, I can build static and shared, but I'm now having a problem compiling test/CachedFactoryTest.

I'll report back after I figure that one out.

[Zak Kipling]

Logged In: YES
user_id=764271
Originator: NO

I get errors there too -- I think it needs Loki:: qualifications on the nested Factory and Key typedefs in CachedFactory.h (lines 665 and 692) in order to be valid according to the stricter scoping rules in 4.3.

[Tom Browder]

[Tom Browder]

Logged In: YES
user_id=861162
Originator: YES

OK, the changes suggested by zak-k worked. Attached is a corrected patch from running svn diff in my updated working directory.

File Added: Loki-lib.patch.gcc4.3.corrected

[Tom Browder]

Logged In: YES
user_id=861162
Originator: YES

I've deleted the original, incorrect patch.

[Tom Browder]

Logged In: YES
user_id=861162
Originator: YES

Is there still an issue with this patch? I believe it should be added to the trunk in time for the next release unless there is something wrong.

Should I file a bug report? I see a previous report and fix for a pre-release version of g++ 4.3. G++ 4.3.1 is now the latest official release and the errors and warnings still exist.

Thanks.