|
| 1 | + Geant4.0.1 Release Notes |
| 2 | + ------------------------ |
| 3 | + |
| 4 | + 14th July 1999 |
| 5 | + |
| 6 | +This version of Geant4 is a "consolidation" of the first public |
| 7 | +release. Its main characteristics are: |
| 8 | + |
| 9 | + o improved reliability and robustness. |
| 10 | + o the possibility of using granular (sub-category) libraries. We |
| 11 | + recommend this mode of installation and use. |
| 12 | + o the possibility of using either the Standard Template Library |
| 13 | + (STL) or RogueWave tools.h++, selectable at installation and use |
| 14 | + time. |
| 15 | + o new low energy electromagnetic processes (first version). |
| 16 | + o extension of nuclear photon-evaporation to include the calculation |
| 17 | + of time to production for gammas. |
| 18 | + o improvements in design of the User Action classes - see below. |
| 19 | + |
| 20 | +Note that the next release, Geant4.1.0, will be offered only in the |
| 21 | +STL version. |
| 22 | + |
| 23 | +This code and some binary libraries are available through our "Source |
| 24 | +Code" Web page - see our <A |
| 25 | +Href="http://wwwinfo.cern.ch/asd/geant4/geant4.html">Geant4 Home |
| 26 | +Page</A>. |
| 27 | + |
| 28 | +Please refer to <A |
| 29 | +Href="http://wwwinfo.cern.ch/asd/geant4/G4UsersDocuments/Overview/html/index.html">Geant4 |
| 30 | +User Documentation</A> for further information about using Geant4. |
| 31 | +Updates to the documentation for Geant4.0.1 are still in preparation |
| 32 | +and are expected to be published within three weeks. |
| 33 | + |
| 34 | + |
| 35 | +Contents |
| 36 | +-------- |
| 37 | + |
| 38 | + 1. Supported and Tested Platforms |
| 39 | + 2. CLHEP 1.4 |
| 40 | + 3. Using the Standard Template Library |
| 41 | + 4. Compiler Specific Problems |
| 42 | + 5. Known Run-Time Problems |
| 43 | + 6. Compilation Warnings |
| 44 | + 7. Known Run-Time Warnings |
| 45 | + 8. Changes of the Signatures of Methods of the User Action Classes |
| 46 | + |
| 47 | + |
| 48 | +1. Supported and Tested Platforms |
| 49 | +--------------------------------- |
| 50 | + |
| 51 | + o SUN Solaris 5.6, C++ 4.2 patch 104631-04. |
| 52 | + o HP 10.20, aCC C++ B3910B A.01.18. |
| 53 | + o Linux 2.0.35, gcc C++ egcs-2.91.60. There is a bug in an egcs |
| 54 | + include file which affects compilation with STL - see comments |
| 55 | + below. This configuration was tested in the RedHat 5.1 |
| 56 | + distribution, but versions of Geant4 have also been compiled |
| 57 | + successfully in Debian and Suse distributions. |
| 58 | + |
| 59 | +Platforms also tested but giving rise to some problems - see below: |
| 60 | + |
| 61 | + o AIX 4.3, xlC compiler. |
| 62 | + o DEC V4.0, cxx C++ V6.1-027. |
| 63 | + o NT 4.0 with Service Pack 4 and Cygnus Tools b20 with: |
| 64 | + Visual C++ 6.0 Service Pack 2 for either STL or RogueWave, or |
| 65 | + Visual C++ 5.0 Service Pack 3 for the RogueWave version only. |
| 66 | + |
| 67 | + |
| 68 | +2. CLHEP 1.4 |
| 69 | +------------ |
| 70 | + |
| 71 | +Geant4.0.1 requires the installation of <A |
| 72 | +href="http://wwwinfo.cern.ch/asd/lhc++/INSTALLATION/clhep.html">CLHEP |
| 73 | +1.4</A>. The following platform specific instructions apply: |
| 74 | + |
| 75 | + DEC: Do *not* use the compiler option "-std strict_ansi". |
| 76 | + |
| 77 | + NT: After configure, edit the file config/CLHEP-x86-cygnus-win32 and |
| 78 | + comment out "#define HEP-USE-STD 1". |
| 79 | + |
| 80 | + |
| 81 | + |
| 82 | +3. Using the Standard Template Library |
| 83 | +-------------------------------------- |
| 84 | + |
| 85 | +The following versions of STL have been tested: |
| 86 | + |
| 87 | + <A href="http://www.objectspace.com/">ObjectSpace</A> STL on: AIX, |
| 88 | + DEC, HP, NT and SUN. |
| 89 | + |
| 90 | + <A href="http://www.stlport.org/">STLPORT</A> STL on: NT |
| 91 | + |
| 92 | + "Native" STL on: Linux. |
| 93 | + |
| 94 | +This is selected at installation/compile time by environment variables |
| 95 | +- see documentation. Be aware that this is the first implementation |
| 96 | +of "The STL Interface" for which the emphasis has been on correct |
| 97 | +behaviour, not on performance. Also, the size of libraries and |
| 98 | +executables is considerably larger. These issues will be addressed in |
| 99 | +the next release. |
| 100 | + |
| 101 | + |
| 102 | +4. Compiler Specific Problems |
| 103 | +----------------------------- |
| 104 | + |
| 105 | + o Linux with egcs-1.1.1 and 1.1.2 |
| 106 | + |
| 107 | + In order to compile with STL it is necessary to edit |
| 108 | + stl_hash_fun.h in /usr/include/g++/ or /usr/local/include/g++/. |
| 109 | + Lines 65-67 must be removed or commented out. These are the |
| 110 | + lines in question: |
| 111 | + |
| 112 | + __STL_TEMPLATE_NULL struct hash<signed char> { |
| 113 | + size_t operator()(unsigned char x) const { return x; } |
| 114 | + }; |
| 115 | + |
| 116 | + o AIX 4.3, xlC compiler. |
| 117 | + |
| 118 | + Geant4 does compile and link on AIX... eventually. The compiler |
| 119 | + is very slow and the executables in debug mode are very large. |
| 120 | + On rsplus at CERN we had to revert to a version of the archiver |
| 121 | + corresponding to AIX 4.1 in order to have templates treated |
| 122 | + correctly. This has limited our ability to test Geant4. |
| 123 | + |
| 124 | + o DEC V4.0, cxx C++ V6.1-027. |
| 125 | + |
| 126 | + We have found cases of mis-compilation of min and max templates |
| 127 | + on DEC. We have traced those which cause serious malfunction |
| 128 | + but there is no guarantee that there are no other cases. |
| 129 | + Detailed inspection of assembler output suggests that this |
| 130 | + problem might not be confined to min and max. This has been |
| 131 | + reported to DEC/Compaq. We understand a new version 6.1A of the |
| 132 | + compiler is on its way. However we cannot guarantee correct |
| 133 | + execution of Geant4 on DEC with the current version of the |
| 134 | + compiler. |
| 135 | + |
| 136 | + o NT |
| 137 | + |
| 138 | + - Note that the G4SYSTEM environment variable used for compiling |
| 139 | + Geant4 on NT has changed from WIN32-VC-NICE to WIN32-VC. |
| 140 | + Please check documentation on setting the environment. |
| 141 | + - Testing was done using the debuggable version only. |
| 142 | + - g3tog4 has not been ported to NT. |
| 143 | + - During the making of dependency files during compilation there |
| 144 | + are repeated warnings ("strstrea.h not found") - this is an |
| 145 | + artefact of using g++ for this purpose. |
| 146 | + - Ignore linker warnings: "conflicts with use of other libs". |
| 147 | + - Ignore linker errors: "unresolved external symbol |
| 148 | + __imp__MessageBoxA@16". |
| 149 | + - There is a known problem reading the data files for neutron |
| 150 | + scattering processes. A patch resolving this problem will |
| 151 | + be issued soon. |
| 152 | + |
| 153 | + |
| 154 | + |
| 155 | +5. Known Run-Time Problems |
| 156 | +-------------------------- |
| 157 | + |
| 158 | + o Reading STEP files on DEC with optimised libraries causes a Memory |
| 159 | + fault - but see notes on DEC above. |
| 160 | + o exampleN04 loops on DEC with optimised libraries - but see notes |
| 161 | + on DEC above. |
| 162 | + o On HP, an executable which uses "integral approach" |
| 163 | + electromagnetic processes - G4IMultipleScattering, |
| 164 | + G4IeIonisation, G4IeBremsstrahlung, G4IeplusAnnihilation, |
| 165 | + G4IhIonisation - has intermittently given Bus error(coredump). |
| 166 | + |
| 167 | + |
| 168 | +6. Compilation Warnings |
| 169 | +----------------------- |
| 170 | + |
| 171 | +There are compilation warnings om some platforms. We do not believe |
| 172 | +that any will lead to incorrect run-time behaviour, but we are working |
| 173 | +on reducing them. |
| 174 | + |
| 175 | + |
| 176 | +7. Known Run-Time Warnings |
| 177 | +-------------------------- |
| 178 | + |
| 179 | +The following messages are written to error output while tracking. We |
| 180 | +believe none give rise to incorrect behaviour. |
| 181 | + |
| 182 | + o G4PropagatorInField::LocateIntersectionPoint: Warning: Integration |
| 183 | + inaccuracy requires an adjustment in the step's endpoint Two |
| 184 | + mid-points are further apart than their curve length difference: |
| 185 | + Dist = xxx curve length = yyy |
| 186 | + o Stepsize underflow in Stepper. |
| 187 | + o Warning G4Navigator::ComputeStep found slightly inaccurate |
| 188 | + position... |
| 189 | + o Warning in G4Navigator::ComputeStep: The Step's starting point has |
| 190 | + moved... |
| 191 | + |
| 192 | + |
| 193 | +8. Changes of the Signatures of Methods of the User Action Classes |
| 194 | +------------------------------------------------------------------ |
| 195 | + |
| 196 | +The signatures of all methods of all of the optional user action |
| 197 | +classes have been changed. Now every method takes a constant pointer |
| 198 | +to an appropriate G4 object. To cope with this change the user code |
| 199 | +must be modified. The examples and documents are updated and the user |
| 200 | +is requested to consult to them. We are sorry for this inconvenience |
| 201 | +but we concluded that this change will provide the user with a better |
| 202 | +and more robust code. |
0 commit comments