From: Arlindo da S. <da...@al...> - 2011-10-12 01:23:55
|
On Tue, Oct 11, 2011 at 6:04 PM, Jennifer Adams <jm...@co...> wrote: > > On Oct 11, 2011, at 5:49 PM, Arlindo da Silva wrote: > > On Tue, Oct 11, 2011 at 5:37 PM, Huddleston, John < > Hud...@ci...> wrote: > >> Hi Arlindo >> >> Sorry for the delayed responses this week. I'm taking leave to take care >> of issues around the place before winter sets in. Today I put up a retaining >> wall with 21 10' 4x4 timber. We live on 35 acres and the four horses are >> a continual challenge. >> > > > Horses? We have a cat and he is already trouble enough... > > >> Tomorrow I have to cut down several trees killed by the pine beetle. We >> have 35 acres and live up at 6850' elevation in the foothills of the rockies. >> It's about 40 miles from Fort Collins or 40 miles from Laramie. >> >> > > It sounds very cool, the kind of life style my wife would not mind... > > >> I received an email from Barbara Jones of the hdfgroup. She is going to >> ask her developers the best way to find the bug. >> >> > > I am pretty sure running hrepack on a sample file would show the > performance issue. No need to bring grads into the picture. > > Barbara (the brain behind the help desk at HDF) is extremely capable. But I > agree it would simplify things if GrADS could be factored out. > > > >> Yes, it is HDF4.2r3. The hdfgroup has gone to a new naming convention so >> I did not want to confuse you. The GrADS version supported by >> the SUPPLIBS works. All newer hdfgroup HDF4 versions have a >> serious performance issue. Do not use HDF4.2r5 (aka hdf-4.2.5) it has >> serious performance issues as does HDF4.2r6 (aka hdf-4.2.6). >> >> Both current HDF5 and NetCDF4 libraries are fine. >> >> > > Is this windows specific or on all platforms? I know for a fact that HDF > 4.2.3 does not build on mac os x 10.6. Perhaps I should hack it until it > builds. > > I have never even tried a more recent version than 4.2r3, but I'd be > willing to do some testing on our CentOS boxes if that would be useful. > > > Jennifer: Have you attempted building HDF 4.2.3 on Snow Leopard? > > I never had a Snow Leopard laptop, I am leapfrogging from 10.5 to 10.7, but > Brian does, and I'm pretty sure I got everything to build for his > environment. The trick I needed was to set CFLAGS to "-arch i386". > But that forces 32 bit only binaries. By default you get universal 32/64 bit binaries. It would be nice to 64 bit binaries going forward, don't you think? This is what we are doing on Linux (although the 32-bit linux builds work everywhere). > I will check on it when I'm in the office tomorrow (the OSX 10.6 server has > to be turned on -- it gets very little use). > Lion is very similar to snow leopard, provided you have the same Xcode version (I have the latest 4.1.1). Here is the error I get compiling 4.2.3 on Snow Leopard (without -arch i386): Making all in hdf Making all in src make all-am gcc -DHAVE_CONFIG_H -I. -I/usr/include/rpc -I/Users/adasilva/src/supplibs-2.2.1/i686-apple-darwin10.8.0/include/ -I/Users/adasilva/src/supplibs-2.2.1/i686-apple-darwin10.8.0/include/ncurses -I/Users/adasilva/src/supplibs-2.2.1/i686-apple-darwin10.8.0/include/curl -I/Users/adasilva/src/supplibs-2.2.1/i686-apple-darwin10.8.0/include/xml2 -I/Users/adasilva/src/supplibs-2.2.1/i686-apple-darwin10.8.0/include/zlib -I/Users/adasilva/src/supplibs-2.2.1/i686-apple-darwin10.8.0/include/szip -I/Users/adasilva/src/supplibs-2.2.1/i686-apple-darwin10.8.0/include/libpng -I/Users/adasilva/src/supplibs-2.2.1/i686-apple-darwin10.8.0/include/jpeg -I/Users/adasilva/src/supplibs-2.2.1/i686-apple-darwin10.8.0/include/jasper -I/Users/adasilva/src/supplibs-2.2.1/i686-apple-darwin10.8.0/include/gd -I/Users/adasilva/src/supplibs-2.2.1/i686-apple-darwin10.8.0/include/freetype2 -I/Users/adasilva/src/supplibs-2.2.1/i686-apple-darwin10.8.0/includepixman-1 -I/Users/adasilva/src/supplibs-2.2.1/i686-apple-darwin10.8.0/include/hdf5 -I/Users/adasilva/src/supplibs-2.2.1/i686-apple-darwin10.8.0/include/zlib -I/Users/adasilva/src/supplibs-2.2.1/i686-apple-darwin10.8.0/include/jpeg -I/Users/adasilva/src/supplibs-2.2.1/i686-apple-darwin10.8.0/include/szlib -DBIG_LONGS -DSWAP -O2 -MT atom.o -MD -MP -MF .deps/atom.Tpo -c -o atom.o atom.c In file included from hdf.h:20, from atom.c:64: hdfi.h:1352: error: expected '=', ',', ';', 'asm' or '__attribute__' before 'you' In file included from hdf.h:20, from atom.c:64: hdfi.h:1355:1: warning: "GOT_MACHINE" redefined hdfi.h:760:1: warning: this is the location of the previous definition hdfi.h:1359:1: warning: "DF_MT" redefined hdfi.h:775:1: warning: this is the location of the previous definition hdfi.h:1361: error: redefinition of typedef 'VOIDP' hdfi.h:779: error: previous declaration of 'VOIDP' was here hdfi.h:1362: error: redefinition of typedef '_fcd' hdfi.h:780: error: previous declaration of '_fcd' was here hdfi.h:1363: error: redefinition of typedef 'char8' hdfi.h:781: error: previous declaration of 'char8' was here hdfi.h:1364: error: redefinition of typedef 'uchar8' hdfi.h:782: error: previous declaration of 'uchar8' was here hdfi.h:1365: error: redefinition of typedef 'int8' hdfi.h:783: error: previous declaration of 'int8' was here hdfi.h:1366: error: redefinition of typedef 'uint8' hdfi.h:784: error: previous declaration of 'uint8' was here hdfi.h:1367: error: redefinition of typedef 'int16' hdfi.h:785: error: previous declaration of 'int16' was here hdfi.h:1368: error: redefinition of typedef 'uint16' hdfi.h:786: error: previous declaration of 'uint16' was here hdfi.h:1369: error: redefinition of typedef 'int32' hdfi.h:787: error: previous declaration of 'int32' was here hdfi.h:1370: error: redefinition of typedef 'uint32' hdfi.h:788: error: previous declaration of 'uint32' was here hdfi.h:1371: error: redefinition of typedef 'intn' hdfi.h:789: error: previous declaration of 'intn' was here hdfi.h:1372: error: redefinition of typedef 'uintn' hdfi.h:790: error: previous declaration of 'uintn' was here hdfi.h:1373: error: redefinition of typedef 'intf' hdfi.h:793: error: previous declaration of 'intf' was here hdfi.h:1374: error: redefinition of typedef 'float32' hdfi.h:791: error: previous declaration of 'float32' was here hdfi.h:1375: error: redefinition of typedef 'float64' hdfi.h:792: error: previous declaration of 'float64' was here hdfi.h:1376: error: conflicting types for 'hdf_pint_t' hdfi.h:794: error: previous declaration of 'hdf_pint_t' was here make[3]: *** [atom.o] Error 1 make[2]: *** [all] Error 2 make[1]: *** [all-recursive] Error 1 make: *** [all-recursive] Error 1 This problem is fixed in 4.2.5. Replacing hdfi.h with the newest version gets it to finish the build, but then it fails the hdf-4 unit tests (make check). Arlindo -- Arlindo da Silva da...@al... |