Cleanup and config change for GPhL-specific file #13
Conversation
dev/gphl/Debian9_qt4/start_mxcube2
Outdated
| # HARDWARE_REPOSITORY_SERVER="/MXCuBE/mxcube/ExampleFiles/HardwareObjects.xml:/MXCuBE/mxcube/repo_test_1" | ||
| # CUSTOM_HARDWARE_OBJECTS_PATH="/MXCuBE/mxcube/extra_HO_test_1:/MXCuBE/mxcube/extra_HO_test_2" | ||
| # CUSTOM_BRICKS_PATH="/MXCuBE/mxcube/extra_brick_test_1:/MXCuBE/mxcube/extra_brick_test_2" | ||
| HARDWARE_REPOSITORY_SERVER="${HO_SERVER}/gphl_mock:/${HO_SERVER}:/${HO_SERVER}/embl_hh_p14" | ||
| CUSTOM_HARDWARE_OBJECTS_PATH="" | ||
| HARDWARE_REPOSITORY_SERVER="${HO_SERVER}/configuration/gphl_mock:/${HO_SERVER}/test/xml-qt:${HO_SERVER}/configuration/alba_xaloc13" |
There was a problem hiding this comment.
This path is ALBA specific, is that the intention ?
There was a problem hiding this comment.
Sort of, yes. This is inside the 'gphl/' directory, which makes it global-phasing specific. Since I test relative to different beamlines I need to have varying specific directories on the lookup path, and it seemed easier to have all the configuring in one place, and checked in, and then change it at need. That way it is visible how things work, in a way it would not be if I relied on some other mechanism, like local environment variables. But if you feel it is not a good idea to have specific locations checked in, even in a gphl/ directory, I can change to a different model
Regrettably the import cleanup (or a small part of it) was required for the bug fix, hence the mixing of the two. The import changes have moved to importing from HardwareRepository. etc. and to gui. etc. in all cases, avoiding direct imports of e.g. sample_centring, abstract.xyz, individual HOs etc. The reason is that importing the same file under two different paths creates two different copies in memory, so that state_x == state_y or setting class attributes gives unpredictable results when the class/instance is not from the same package in different cases.
Regrettably the import cleanup (or a small part of it) was required for the bug fix, hence the mixing of the two. The import changes have moved to importing from HardwareRepository. etc. and to gui. etc. in all cases, avoiding direct imports of e.g. sample_centring, abstract.xyz, individual HOs etc. The reason is that importing the same file under two different paths creates two different copies in memory, so that state_x == state_y or setting class attributes gives unpredictable results when the class/instance is not from the same package in different cases.
Added dose_budget and transmission calculation, multiple characterisation strategies, ...
Debugging still incomplete
|
I will see you later today, if I forget, remind me to ask you about this PR :) |
|
@marcus-oscarsson The actual changes here are in two parts:
Can we get this merged as it is, or what would I need to do tro get it movede forward? |
(that do not rquire java code changes)
This reverts commit 2ec3319
NB some refactoring for consistency NB Removed cp parameter for collect (for now)
Added docker for Ubuntu:focal, Python3, Qt5,