/* versionpatch.pl Version patch code. Used by garpModel->versionPatch when a model is loaded that uses an older modelDefinitionVersion than the current one. Loaded only when needed. This file defined pl_doVersionPatch(Model, CurrentVersion, ModelVersion) to get Model from ModelVersion to CurrentVersion. For every step (from version 0 to 1, 1 to 2 etc) is has code, so we can patch any old version. 8-12-2005 Jelmer Jellema (Spin in het Web) Part of GARP3 See copyright notice %gp3 1.4: Due to heavy changes in data structure, we no longer support models with a version below 15. Users should load and save their version in Garp3 version 1.3.8 before using 1.4. %gp3 1.4: We made some changes. pl_doVersionPatch(M,CurrentVersion,LowerVersion) is now a main call. It first executes a series of calls looping from CurrentVersion back to LowerVersion + 1. These new calls pl_versionPatch_data(M,Version), are meant to set newly added variables. It was invented to add the translator to the model in version 15. This is needed, because the version patch for version 8 (!!) initialises data that in the current code needs this translator. After this, the old pl_versionPatch(M,OldVersion) is still used to loop from LowerVersion to CurrentVersion. So: - use pl_versionPatch_data(M,Version) to define and ininitialise variables and other data that were added in Version. - use pl_versionPatch(M,OldVersion) to manipulate old data (probably in @old_slots) to get to the new version. All elements initialises in pl_versionPatch_addElements are there... */ :-discontiguous pl_versionPatch/2, pl_versionPatch_addElements/2. % pl_doVersionPatch(M,CurrentVersion,LowerVersion):- pl_doVersionPatch_data(M,CurrentVersion,LowerVersion), pl_doVersionPatch_patches(M,CurrentVersion,LowerVersion), ( LowerVersion < CurrentVersion % patch was done -> get(@app?version, value, Version), format(atom(String), 'Model definition was updated from version ~w to ~w.\n\nPlease save your model if you intend to keep working in GARP ~w.', [LowerVersion, CurrentVersion, Version]), Dlg *= notificationDlg(@app?mainMenu, notification, String,@on), Dlg<<-confirm_centered(@app?mainMenu?frame?area?center), Dlg->>destroy %@app->>setModelChanged(@on) % FL patching is a change, right? But towards the user it's a bit too agressive... ; true % no patch: LC = CV (or greater) ). pl_doVersionPatch_data(M,CurrentVersion,LowerVersion):- %gp3 1.4. Loop back from CurrentVersion to LowerVersion + 1, initialising any data added in CurrentVersion LowerVersion < CurrentVersion,!, % This version patch rolls back over version numbers, so the patch does not necessary exist. Sort of scary. % For this reason, the patch can fail, requiring the fail to be ignored ignore(pl_versionPatch_data(M,CurrentVersion)), PrevVersion is CurrentVersion - 1, pl_doVersionPatch_data(M,PrevVersion,LowerVersion). % pl_doVersionPatch_data(_,_,_). %done. %% pl_doVersionPatch_patches(M,CurrentVersion,LowerVersion):- %gp3 %check if there is a patch for models being at a lower definition level LowerVersion < CurrentVersion,!, ( pl_versionPatch(M,LowerVersion), ! ; format(atom(String), 'Version data patch from ~w to ~w failed! Do NOT save your model, as it will potentially cause corruption. Please report this issue to the developers, and send them your model.', [LowerVersion, CurrentVersion]), Dlg *= notificationDlg(@app?mainMenu, notification, String,@on), Dlg<<-confirm_centered(@app?mainMenu?frame?area?center), Dlg->>destroy, fail ), NextVersion is LowerVersion + 1, pl_doVersionPatch_patches(M,CurrentVersion,NextVersion). % pl_doVersionPatch_patches(_,_,_). %done %% /* pl_versionPatch_data(M,NewVersion) %gp3 1.4: This one is called for every version level starting at current and looping back %use it to initialise data that is new in NewVersion. This way pl_versionPatch calls can asume %all current *code* runs fine, because all variables are there. Without it, a patch at level 8 can fail %because a class definition uses data added at level 9. */ pl_versionPatch(M,20):- % From 20 to 21 added yet another preference FL RP = M<<-runPrefs, RP->>append(solve_extra_combinations,@on), @pce->>write_ln('20 --> 21: Added 1 new simulation preference.'). pl_versionPatch(M,19):- % From 19 to 20 added yet another preference FL RP = M<<-runPrefs, RP->>append(full_branching_derivative_terminations,@off), @pce->>write_ln('19 --> 20: Added 1 new simulation preference.'). pl_versionPatch(M,18):- % From 18 to 19 added yet another preference FL RP = M<<-runPrefs, RP->>append(fast_path,@off), @pce->>write_ln('18 --> 19: Added 1 new simulation preference.'). pl_versionPatch(M,17):- % From 17 to 18 we another preference FL RP = M<<-runPrefs, RP->>append(equal_intervals,@off), @pce->>write_ln('17 --> 18: Added 1 new simulation preference.'). pl_versionPatch(M,16):- % From 16 to 17 we added some preferences and removed some... FL RP = M<<-runPrefs, RP->>append(equal_qspace_points,@off), RP->>append(epsilon_derivative_constraints,@on), RP->>append(reasoning_assumptions,@on), RP->>append(assume_conditional_derivatives,@on), RP->>append(max_depth, 0), RP->>append(second_order_derivatives,@on), RP->>append(derivative_terminations,@on), RP->>append(second_order_continuity,@on), RP->>append(order_epsilon_last,@off), RP->>append(value_branching,@off), ignore(( RP->>delete(no_subsumption_specify_all), RP->>delete(use_old_transition_rules), RP->>delete(no_analyse_zero), RP->>delete(allow_d_assumptions_in_reclassifying_mfs) )), @pce->>write_ln('16 --> 17: Added 10 new simulation preferences, and removed 4 old ones.'). pl_versionPatch_data(M,16):- @pce->>write_ln('16: translator added'), M->>translator(new(translator)). %% /* pl_versionPatch(M,OldVersion) gp3 0.3 rewrite of old versionpatch this one is called for every version level between the current one and the model's one pl_versionPatch(M,1) means: patch M so that it current version (version 1) goes to the next one (version 2) so we can implement patches for any version wanted Version 0 is the code we used to patch stuff before explicit versioning So here we have to do checks about the content */ pl_versionPatch(M,15):- %gp3 1.4 (JJ). Add translation support % add model fragments in the hierarchy to @model?modelFragments if they are not there already reRegisterModelFragments(@model), % JL: There are sometimes datastructure remants of an old Sketch in the abstractEntities that should be removed removeOldSketchDatastructures(@model), %a lot of conversion from old char_array values to translatables, using the @old_slots hash %(see object_extensions.pl: object->convert_old_slot) @pce->>write_ln('15 --> 16: Add translation support'), TR = M<<-translator, %added through pl_versionPatch_data above. %%a lot of objectsd have name and remarks translatable, we do one loop %%sketch stuff is also in subclasses of abstractEntity, it appears (did I allready ask why?) Obs *= chain, chain(M?modelFragments,M?abstractEntities,M?assumptions,M?configurationDefinitions,M?attributeDefinitions,M?quantityDefinitions,M?quantitySpaces,M?dqs)->>for_all( if(->>(@arg1,instance_of,chain), ->>(Obs,merge,@arg1), ->>(Obs,append,@arg1) ) ), Obs->>for_all( and( ->>(@arg1,name_translatable,?(TR,getTranslatable, unique)), ->>(@arg1,name,?(?(@old_slots,member,@arg1?object_reference),member,name)), ->>(@pce,write_ln,@arg1,?(@arg1,name)), ->>(@arg1,remarks_translatable,?(TR,getTranslatable, empty)), ->>(@arg1,remarks,?(?(@old_slots,member,@arg1?object_reference),member,remarks)) )), %%Values for attributeDefinitions M?attributeDefinitions->>for_all( ->>(?(@arg1,slot,values),for_all, %use slot, because it is wrapped thru a copy action and( ->>(@arg1,valueName_translatable,?(TR,getTranslatable,unique)), ->>(@arg1,valueName,?(?(@old_slots,member,@arg1?object_reference),member,valueName)), ->>(@pce,write_ln,'Attrib value:',@arg1?valueName) ) ) ), %%values for quantityspaces QS = M?quantitySpaces<<-copy, QS->>append(M?dqs), QS->>for_all( ->>(?(@arg1,slot,values),for_all, and( ->>(@arg1,valueName_translatable,?(TR,getTranslatable,unique)), ->>(@arg1,valueName,?(?(@old_slots,member,@arg1?object_reference),member,valueName)), %value Zero is nontranslatable: if(->>(@arg1?valueName,equal,'Zero'),->>(@arg1,setTranslatable,@off)), ->>(@pce,write_ln,'QS value:',@arg1?valueName) ) ) ), %2 special quantityspaces are nontranslatable (name and remarks allready set above) chain(M?dqs, M?mzpqs)->>for_all( and( ->>(@arg1?name_translatable,setNonTranslatable), ->>(@arg1?remarks_translatable,setNonTranslatable), ->>(?(@arg1,slot,values),for_all,->>(@arg1,setTranslatable,@off)) %again, we use the slot, because otherwise we ll get a copy ) ), %model fragment elements M?modelFragments->>for_all( and( ->>(?(@arg1?elements,merge,@arg1?parentReferences),for_all, and( ->>(@pce,write,@arg1), if(?(@arg1,slot,name_translatable), %has name_translatable and( ->>(@arg1,name_translatable,?(TR,getTranslatable,unique)), ->>(@arg1,name,?(?(@old_slots,member,@arg1?object_reference),member,name)), ->>(@pce,write,' name',@arg1?name) ) ), if(?(@arg1,slot,remarks_translatable), %has remarks_translatable and( ->>(@arg1,remarks_translatable,?(TR,getTranslatable,empty)), ->>(@arg1,remarks,?(?(@old_slots,member,@arg1?object_reference),member,remarks)), ->>(@pce,write,' remarks') ) ), %all other elements containing valueReference objects %(@arg3 is the slot name, @arg4 is the object in the slot) ->>(@arg1,for_slot_reference, if(and(@arg2 == 'slot', ->>(@arg4,instance_of,valueReference)), and( ->>(@arg4,valueName_translatable,?(TR,getTranslatable,unique)), ->>(@arg4,valueName,?(?(@old_slots,member,@arg4?object_reference),member,valueName)), ->>(@pce,write,' valueReference:', @arg3,'=',@arg4?valueName) ) ) ), ->>(@pce,write_ln) ) ) )), %special for top nodes M?topStaticFragment?name_translatable->>setNonTranslatable, M?topStaticFragment?remarks_translatable->>setNonTranslatable, M?topProcessFragment?name_translatable->>setNonTranslatable, M?topProcessFragment?remarks_translatable->>setNonTranslatable, M?topAgentFragment?name_translatable->>setNonTranslatable, M?topAgentFragment?remarks_translatable->>setNonTranslatable, @pce->>write_ln('15 --> 16: Finished adding translation support'). %%