-
Notifications
You must be signed in to change notification settings - Fork 16
Conversation
…tgwbse calculator: - full dftgwbse.xml file in share folder - calculator passes dftpackage options and gwbse options explicitly through, instead of the xml files - xml file reading only on calculator level Still not good: - basisset, functional definitions are repeating - some tags are superfluous - xtp_tools still needs to be passed an user option file on the command line, which can be empty though
DeepCode failed to analyze this pull requestSomething went wrong despite trying multiple times, sorry about that. |
- basisset, auxbasisset, functional now given on base level - dftgwbse.cc then transfers this into to both DFT and GWBSE options - reporting removed, was not used at all - archive file name now derived from XYZ file name - XML output summary name now derived from XYZ file name
I am not sure if all options should have a default, i.e. xyz file. Why do you want to default everything? I mean we can also put the xyz input into the xml file, if we want to reduce the number of files |
The xyz input default does not hurt anyone. Adding the coordinates into the xml file is something I don't like. What people will always have is a coordinate file in most likely that format. Asking them to add xml tags around it is just going to be annoying for people. And why defaulting everything? Just consider the minimum input file needed for |
so orca does not default on the |
Yes. We can "undefault" the xyz file name of course since a calculator as defined way down in
If we require xyzfile to be specified, at least it will make sense to require the options file:
|
removed mo and log file options from gwbse engine
Felipe defaults xml
Merge after votca/votca#263 |
It makes one test fail:
|
Codecov Report
@@ Coverage Diff @@
## master #413 +/- ##
======================================
Coverage 64.3% 64.3%
======================================
Files 281 281
Lines 23663 23663
======================================
Hits 15233 15233
Misses 8430 8430
Continue to review full report at Codecov.
|
@JensWehner please review! Do we need tests? |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I have two major concerns with this PR still
a) I do not get why we update the defaults with user options?
b) I am not sure, if it is a good idea to default everything.
I think that updating the defaults with the user input is the normal behavior that I would expect of a general simulation package. As an author of a simulation package, you don't want all your users to enter every single variable for configuring all the details of your simulation neither force them to make a copy of the whole options file and manually update the sections that the user are interested in. Instead, most simulation packages allow the user to provide minimal input that is subsequently merged with the whole set of configuration variables.
Put yourself on the user's shoes, you would like to try this amazing XTP package to do a DFT-GWBSE simulation, if instead of having a very simple input interface (that can be modified for expert users) you force your users to enter every single detail of the simulation, the potential users would run away ;) |
Okay I agree with a). Concerning b) I agree that most things can be defaulted, but inputfilenames should not be defaulted in my opinion. |
How do you envision the input/defaults handling?
How shall we then handle the input filenames? |
I thought we do it the same way as before. I agree that 90% of all options have a sensible default, but I do not think that the coordinates file, or the output file have that. |
Apart from eanalyze and ianalzye I cannot see how you would default all options of a calculator. |
@JensWehner please review. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I still do not think defaulting all options is a good idea, but the PR has a lot more good than bad things in it.
Very first attempt at getting new default xml file behavior, testing on dftgwbse calculator:
Still not good:
Fix #319