Command line functionality

[ms609]

The "Batch mode" dialog seems to allow for multiple replicates with identical run settings, but I can't see a way to programmatically run multiple iterations with different settings: for instance, to run a hundred simulations in which each has a different rate parameter – which is what I would (perhaps wrongly?) understand from the term "batch mode".

Does such a functionality exist? Perhaps what I'm really looking for is just a way to launch TREvoSim from the command line with parameters specified manually (or by a script); or to launch an instance of TREvoSim that runs automatically from a given settings save file.

[RussellGarwood]

Thanks for these thoughts. My intention for the batch mode was to run replicates with the same settings - this is the same in REvoSim, so is consistent terminology across our packages.

At the moment, no functionality such as that you request exists. Long term I plan to faciltate this with a headless vanilla C++ version (see comment in #3 ), however, I won't be able to achieve that for this release. I may be able to achieve a command line version that loads a settings file then runs once for this release, time constraints allowing. If I do I will post here, and if not I will leave this issue open until one is in place.

[ms609]

That makes sense, thanks – it certainly isn't necessary in the context of the review, but it will be a useful addition when it comes.

[RussellGarwood]

@ms609 I've taken advantage of a fairly quiet moment in the Museum fur Naturkunde to implement something along these lines, and test it on Linux - see above push. If you launch the software from the command line with the switch -o or --open, pointing towards a settings file, it should load that file, and then run the simulation to completion before quitting the software.

There are a few things that could trip it up with Windows, but I am not able to test directly right now. Any feedback from you or @alanspencer would be appreciated when/if either of you gets the chance.

• In theory, the parser should be fine with paths however you write them - i.e. / or \ as separators. It would be good to know for sure if this is the case
• Windows operates differently to Unix systems in terms of messaging, and when you run this it should provide informative messages to the console on error, or indeed, on completion of a run. I used qInfo to provide these, which may be ported to a debugger if Qt thinks this is not a console application (but presumably if you launch it from the console qt will work out that it is). It'd be great to know if the messages make it through to console when you do a run
• Exiting the applicaiton in any graceful way seems not to work, but the brute force approach I've used here works fine on Linux. It'd be good to know if it provides an error message on exit on windows

I will add this to the documentation once it is clear whether it is working OK on all systems!