Closed maggie-klureza closed 9 months ago
The nature of this log works as intended, except that %d
is not rendering properly here. That should be an integer number (that is the usual string format specifier in Python). I believe I've seen this issue when using multiprocessing and without multiprocessing it renders fine -- that suggests it's a DIALS-side issue. I'll confirm this and forward to the DIALS devs if so
Yes, the part where %d
doesn't render is what I meant by not containing useful information - the fact that it's an integer number doesn't tell me much without knowing what that integer is. @dennisbrookner said he'd seen this as an issue with laue-dials previously (and it sounded like not in the context of non-laue DIALS?).
. @dennisbrookner said he'd seen this as an issue with laue-dials previously (and it sounded like not in the context of non-laue DIALS?).
Yes this is correct; I can confirm that I have seen the %d
behavior in laue-dials
and its "laue-index-assigner
" predecessor. I don't recall ever seeing this in regular dials
, but I don't know that I've really looked for it
@maggie-klureza I was just looking at this on my end and I cannot reproduce it. It's possible this was a DIALS-side issue with multiprocessing and they've fixed it recently. Does this issue still occur for you?
I just made a new conda environment with fresh installations of dials and laue-dials ( conda install -c conda-forge dials, pip install laue-dials
, as per the tutorial notebook) and reran the import and spotfinding in a new directory. I still get the same issue - spotfinding log is full of %d
Which version of DIALS + laue-dials are you using? You can use pip list
to get a complete set of packages with version numbers.
Looks like installing the packages as described above (aka as directed in the tutorial notebook) this morning led me to dials=3.17.0
and laue-dials=0.3
. If that's an insufficiently recent version of dials, then the tutorial instructions on setting up a conda environment with the necessary packages should be revised.
This issue is solved by this commit. DIALS has removed the force-2d
parameter and the nonexistent parameter was causing this issue. laue-dials
now has also removed the parameter.
The log file written by laue_dials.find_spots (which seems to be laue_output.log rather than output.log - see other issue) contains a lot of text that is formatted like a spot finding log but doesn't actually contain useable information:
This continues for many more iterations. It seems like the spot finding does proceed successfully, but it's hard to evaluate much from the log!