Not 100% sure on this, but it looks like the simulation runs for conc 0 and fails on conc 0.001, but the error gets written to chaste_errors_0.txt, not chaste_errors_0.001.txt!

--model 1 --pacing-freq 1.0 --pacing-max-time 5 --pic50-nal 1 --hill-nal 1 --saturation-nal 0 --plasma-conc-high 100 --plasma-conc-low 0 --plasma-conc-count 4 --plasma-conc-logscale true

When doing the following I seem to end up with wget output from EXPECT0 going to stderr, whereas I'd have hoped for it to go to stdout.

--model 8 --pacing-freq 1.0 --pacing-max-time 5 --pic50-herg 1 --hill-herg 1 --saturation-herg 0 --pic50-spread-herg 0.21 --plasma-conc-high 100 --plasma-conc-low 0 --plasma-conc-count 4 --plasma-conc-logscale true  --credible-intervals 2> stderr

Hi @gary, I'm noticing that there's an inconsistency between data being written to voltage_results.dat and q_net.txt.
In voltage_results.dat, if there are "NoActionPotential_X" values at concentrations, then it's appearing as :
23.1013 79.5776 27.5461 512.202 621.956 114.61 81.1131 NoActionPotential_1 NoActionPotential_1 NoActionPotential_1 NoActionPotential_1 NoActionPotential_1
In q_net.txt, it's just :
23.1013 -0.0296737
i.e. for q_net.txt I'd be expecting 81.1131 NoActionPotential_1
Observed with: --model 8 --pacing-freq 0.5 --pacing-max-time 5 --pic50-cal 5 --hill-cal 1 --saturation-cal 0 --pic50-herg 5 --hill-herg 1 --saturation-herg 0 --pic50-iks 6 --hill-iks 1 --saturation-iks 0 --pic50-na 6 --hill-na 1 --saturation-na 0 --plasma-conc-high 1000 --plasma-conc-low 0 --plasma-conc-count 10 --plasma-conc-logscale true

It would be nice that ApPredict still work with older installations of chaste that are still using c++11 and pycml, I would like to make the following changes if there are no objections?

diff --git a/SConscript b/SConscript
index e111a1d..e7125c8 100644
--- a/SConscript
+++ b/SConscript
@@ -48,6 +48,9 @@ chaste_libs_used = ['heart']
 # Change some flags just for this project
 env = SConsTools.CloneEnv(env)
 env['Chaste_CODEGEN_EXTRA_ARGS'] = ['--use-model-factory']
+env['PYCML_EXTRA_ARGS'] = ['--expose-annotated-variables']
+
+

 # Do the build magic
 result = SConsTools.DoProjectSConscript(project_name, chaste_libs_used, globals())
diff --git a/src/fortests/ModelFactory.cpp b/src/fortests/ModelFactory.cpp
index 02a3e2f..2b90129 100644
--- a/src/fortests/ModelFactory.cpp
+++ b/src/fortests/ModelFactory.cpp
@@ -41,7 +41,7 @@ ModelFactory::TModelMapping ModelFactory::getModelRegistry()
 {
     if (ModelFactory::mpModelRegistry == nullptr)
     {
-        ModelFactory::mpModelRegistry = std::make_unique<std::map<std::pair<std::string, std::string>, ModelFactory::TCreateMethod>>();
+        ModelFactory::mpModelRegistry = std::make_shared<std::map<std::pair<std::string, std::string>, ModelFactory::TCreateMethod> >();
     }
     return ModelFactory::mpModelRegistry;
 }

Would just do this, but probably will break the github actions?

I can invoke ApPredict with nonsensical args like --fish chips and it ignores them. Would be helpful to to have something like ...

• simply fail on invalid args, or,
• add another arg, e.g. --strict-args {true|false}, which defaults to false if not supplied (so ApPredict behaves as it's always done), but if --strict-args true it'll fail to run on bad args.

Things to consider:

• Create a new branch of www.github.com/Chaste/cellml which contains only the files needed by ApPredict, so that compilation is faster and easier.
• By setting some global bash variable specify a folder of CellML files which we can use as additional --model s, either at compile time (by treating it as a source folder) or at run time (by reading contents and dynamically loading them).

I'm seeing the following (seemingly inconsequential, as the simulation seems to complete) "warning" appearing in ApPredict https://github.com/Chaste/ApPredict/releases/tag/v2021.1 stderr in a container.

--------------------------------------------------------------------------
[[60025,1],0]: A high-performance Open MPI point-to-point messaging module
was unable to find any relevant network interfaces:

Module: OpenFabrics (openib)
  Host: 466e13a9963f

Another transport will be used instead, although this may result in
lower performance.

NOTE: You can disable this warning by setting the MCA parameter
btl_base_warn_component_unused to 0.
--------------------------------------------------------------------------

Simulation was run using :

* model = shannon_wang_puglisi_weber_bers_2004_model_updated
* na: no drug effect
* cal: no drug effect
* herg IC50s = 100  uM, Hills = 1, Saturation levels = 0  %.
* iks: no drug effect
* ik1: no drug effect
* ito: no drug effect
* nal: no drug effect
* max free plasma concentration = 100 uM
* min free plasma concentration = 0 uM
* number of plasma concentrations = 13

It'd be an improvement if this wasn't appearing in the logs, so are there any suggestions for adjusting https://github.com/CardiacModelling/appredict-docker/blob/master/appredict-no-emulators/Dockerfile#L40-L47 to include that btl_base_warn_component_unused instruction?

Description
FetchContent currently skips downloading CellML files if configuration has already been done once. This fix was added in #26 and prevents FetchContent from throwing errors when it can't overwrite existing files, which can happen when cmake is called twice or ccmake is used.

The current behaviour means that CellML files which have been accidentally removed after the inital configuration will not be replaced when cmake is called again.

Steps To Reproduce

cd /path/to/Chaste/build
cmake ..
rm /path/to/Chaste/projects/ApPredict/src/cellml
cmake ..
make ApPredict

Error

[ 82%] Building CXX object projects/ApPredict/CMakeFiles/chaste_project_ApPredict.dir/src/stats/BayesianInferer.cpp.o
/home/runner/_work/ApPredict/ApPredict/Chaste/projects/ApPredict/src/single_cell/CipaQNetCalculator.cpp:42:10: fatal error: ohara_rudy_cipa_v1_2017Cvode.hpp: No such file or directory
   42 | #include "ohara_rudy_cipa_v1_2017Cvode.hpp"
      |          ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
make[3]: *** [projects/ApPredict/CMakeFiles/chaste_project_ApPredict.dir/build.make:258: projects/ApPredict/CMakeFiles/chaste_project_ApPredict.dir/src/single_cell/CipaQNetCalculator.cpp.o] Error 1
make[3]: *** Waiting for unfinished jobs....
[ 82%] Building CXX object projects/ApPredict/CMakeFiles/chaste_project_ApPredict.dir/src/stats/LinearDiscriminantAnalysis.cpp.o
make[2]: *** [CMakeFiles/Makefile2:22324: projects/ApPredict/CMakeFiles/chaste_project_ApPredict.dir/all] Error 2
make[1]: *** [CMakeFiles/Makefile2:2327: CMakeFiles/project_ApPredict.dir/rule] Error 2
make: *** [Makefile:314: project_ApPredict] Error 2

To Do

• Try to force overwrites (delete and re-fetch), or check for missing files and fetch only those.

It goes into the provenance file, but would be nice to have it on std::out along with Chaste version too.

Which just prints the commit hashes for Chaste and ApPredict that the executable was compiled with.

These are currently printed to std::out, but be nice to put them in a special call.

Maybe also print the dependency versions (TestChasteBuildInfo.hpp style).