ctffind_runner better error message

[huwjenkins]

Possible fix for issues like #1167? When CTFFIND fails to fit CTF parameters for all input micrographs (i.e. MDctf is empty) exit with an error rather than failing later.

[huwjenkins]

Note I only tested for the SPA case but based on the contents of the file written here:

relion/src/ctffind_runner.cpp

Lines 487 to 490 in 05c903d

	else
	{
	obsModel.save(MDctf, fn_out + "micrographs_ctf.star", "micrographs");
	}

I guess for tomo the underlying issue is the same

[biochem-fan]

@huwjenkins Thank you very much.
Can you also add LC_ALL=C to the CTFFIND caller script written by this wrapper?

[huwjenkins]

@biochem-fan is there a syntax that works for all possible shells? The user can choose this via RELION_SHELL environment variable:

relion/src/ctffind_runner.cpp

Lines 104 to 108 in 05c903d

	fn_shell = "/bin/sh";
	char *shell_name;
	shell_name = getenv("RELION_SHELL");
	if (shell_name != NULL)
	fn_shell = (std::string)shell_name;

[biochem-fan]

The user can choose this via RELION_SHELL environment variable

Oh, that is a pain.

Does env LC_ALL=C ctffind ... work?

[huwjenkins]

It worked when RELION_SHELL was unset but when I put

export RELION_SHELL=/bin/csh

in my SLURM submission script then although CTFFIND doesn't segfault, it doesn't run:

        **   Welcome to Ctffind   **

            Version : 4.1.14
           Compiled : May  8 2020
               Mode : Scripted

Input image file name                              :
Error: File does not exist, please provide an existing file!


 Error: Running as script, and answer is not recognized...

In fact the same behaviour occurs without my changes and using a version of CTFFIND I know works.

[huwjenkins]

export RELION_SHELL="/bin/csh -f"

worked, so I think it was a local issue. I've added the suggested env LC_ALL=C

[biochem-fan]

Thank you very much for your contribution. I merged your patches.

[huwjenkins]

Thanks! FWIW CTFFind 5.0.2 from https://grigoriefflab.umassmed.edu doesn't need the LC_ALL=C workaround so this change shouldn't be propagated to a future CtffindRunner::executeCtffind5()