search

m3g / packmol

Packmol - Initial configurations for molecular dynamics simulations
http://m3g.github.io/packmol
MIT License
214 stars 48 forks source link

regression in packmol-memgen 2024.2.9 #68

Closed xiki-tempula closed 2 months ago

xiki-tempula commented 2 months ago

I have a system where I could use packmol to pack using the packmol bundled in ambertools 23.3 (packmol-memgen=2023.2.24) but not with the packmol bundled in ambertools 23.6 (packmol-memgen=2024.2.9)

Here is how to reproduce the regression.

conda create -n ambertools23.3 -c conda-forge ambertools=23.3
conda activate ambertools23.3
packmol < packmol.inp

Runs fine

conda create -n ambertools23.6 -c conda-forge ambertools=23.6
conda activate ambertools23.6
packmol < packmol.inp

Gives 

  ENDED WITHOUT PERFECT PACKING: 
  The output file:

    bilayer_system.pdb

  contains the best solution found. 

  Very likely, if the input data was correct, 
  it is a reasonable starting configuration.
  Check commentaries above for more details.

Archive.zip

lmiq commented 2 months ago

Are the "packmol.inp" files identical?

xiki-tempula commented 2 months ago

They are identical. I just realised that they give the same result. It is just that the old one return exit code 0 despite the "ENDED WITHOUT PERFECT PACKING" message, while the new one returns an exit code 137.

lmiq commented 2 months ago

Oh, ok. So the new one is the correct one. The proper exit codes were added in newer versions. Probably the solution obtained is good enough, if people where using it in packmol-memgen (I'm not the author nor maintainer of that, so I cannot say much about its choices). How to handle these situations should be something to be chosen in that package.

xiki-tempula commented 2 months ago

Thanks for the explanation.