https://github.com/aiidateam/aiida_core

sort by:
Revision Author Date Message Commit Date
3ba9e1e Merge pull request #938 from sphuber/release_v0.10.0rc4 Release v0.10.0 20 November 2017, 13:16:57 UTC
9a15b1c Release v0.10.0 20 November 2017, 13:06:47 UTC
0bf5c91 Merge pull request #935 from AntimoMarrazzo/fix_cmdline_plugins_tuple Fixed bug in cmdline calculation plugins for tuple of input params 20 November 2017, 09:25:18 UTC
1d4e21d Fixed bug in cmdline calculation plugins for tuple of input params 17 November 2017, 17:25:47 UTC
69927b5 Merge pull request #920 from merkys/with_tcod_reference Adding a reference to the recent AiiDA/TCOD paper 17 November 2017, 11:35:09 UTC
cd8d3cd Merge branch 'release_v0.10.0rc4' into with_tcod_reference 17 November 2017, 11:25:14 UTC
75a1da2 Adding a page number. 17 November 2017, 11:22:15 UTC
04c56bf Adding journal volume to the citation of AiiDA/TCOD paper, page numbers are probably not assigned yet. 17 November 2017, 11:18:33 UTC
7354d43 Merge pull request #926 from lekah/remove_qe_tools_dependencies_2 Remove qe tools dependencies 2 17 November 2017, 11:15:44 UTC
8426727 Merge branch 'release_v0.10.0rc4' into remove_qe_tools_dependencies_2 17 November 2017, 11:03:58 UTC
3ef8bdd Merge pull request #928 from sphuber/fix_927_tcod_exporter_plugin_logic Fix the logic for loading the correct tcod exporter plugin 17 November 2017, 11:02:20 UTC
8884c2c Fix the logic for loading the correct tcod exporter plugin 17 November 2017, 10:29:30 UTC
c86e9f1 Merge branch 'release_v0.10.0rc4' into remove_qe_tools_dependencies_2 17 November 2017, 08:37:00 UTC
b92afd5 Merge pull request #921 from nmounet/fix_837_visualization_with_ase_pygtk_workaround_in_doc Adding pygtk workaround in troubleshooting section of the installation doc 16 November 2017, 11:03:39 UTC
4b13ba6 Adding pygtk workaround in troubleshooting section of the installation doc 16 November 2017, 10:38:53 UTC
986b1ef Adding a reference to recent AiiDA/TCOD paper. 16 November 2017, 08:37:20 UTC
1ef1135 Merge pull request #918 from giovannipizzi/fix_not_exiting_in_upf_command Missing sys.exit(1) 15 November 2017, 16:16:25 UTC
eeb4e3a Merge branch 'release_v0.10.0rc4' into fix_not_exiting_in_upf_command 15 November 2017, 15:38:50 UTC
6b404a4 Merge pull request #917 from AntimoMarrazzo/fix_quotes_write_fortran_input Fixed writing of quotes 15 November 2017, 15:38:39 UTC
bcae7f3 Missing sys.exit(1) 15 November 2017, 15:16:56 UTC
15e9cec Fixed writing of quotes for fortran input files (to support Wannier90 input) 15 November 2017, 13:21:23 UTC
07f00ff Merge pull request #913 from aiidateam/release_v0.10.0rc3 Release v0.10.0rc3 14 November 2017, 12:47:29 UTC
751313d Merge pull request #914 from sphuber/release_v0.10.0rc3 Add an entrypoint 'aiida.tests' that can be used by plugins to add tests 13 November 2017, 16:08:37 UTC
e083b26 Add an entrypoint 'aiida.tests' that can be used by plugins to add tests This entrypoint is scanned by the logic in verdi devel tests and will make sure to add all the tests in this entry points in the tests that are to be run. 13 November 2017, 15:40:27 UTC
fda2a4d Release v0.10.0rc3 * Update version number * Update CHANGELOG.md * Check and update all file headers 13 November 2017, 15:00:33 UTC
30698bd Changed QeInputFile to PwInputfile, because otherwise namelist is not parsed 10 November 2017, 16:51:25 UTC
89e8c00 For #991 we moved functionality to a new repository qe-tools. This is now in the optional requirements. also update command-line for verdi data structure import using the qeinputparser 10 November 2017, 14:53:09 UTC
9480883 Change the expectation value for failed process in daemon tests In the previous commit, logic was added to the process loading code of the daemon that if loading a process from a pickle excepts the associated WorkCalculation node is properly set to failed and sealed. The test_create_fail tested that the node of the failed process was not sealed, which was the old behavior. With the new changes the node is in fact expected to be sealed, so we change the expected outcome of the test 10 November 2017, 13:50:51 UTC
7e7bfcb When a pickle is corrupted, attempt to neatly seal the node If the daemon attempts to load a Process from a pickle that excepts, because it is corrupted, we should at least try to load the corresponding node, set it to failed and seal it, such that any parent processes that might have called it, get a proper signal that the process they were waiting on has finished, albeit unsuccessfully. The parent processes will then not linger in a forever unfinished state, but will be taken out of limbo. To this end we try to load the node, by getting the pk from the pickle filename. Note that this is not very robust as if at some point the format of the pickle name changes and it no longer is {node_ppk}.pickle than there is no way to retrieve the corresponding node. 10 November 2017, 13:50:51 UTC
cac692c Ensure that corrupt pickle get moved to failed directory When the loading of a pickle fails, most likely due to it being corrupted, it might be useful to inspect its state for debugging. Therefore, instead of deleting it, we move it to the failed pickle directory. I also make sure that all pickle directories exists by calling _ensure_directory on the passed paths upon construction of the Persistence instance 10 November 2017, 13:50:51 UTC
cc9d06b Merge pull request #908 from sphuber/fix_907_add_exception_to_workcalculation_report Trigger the DbLogHandler for exceptions in WorkChains run by daemon 10 November 2017, 13:45:07 UTC
ac9d226 Trigger the DbLogHandler for exceptions in WorkChains run by daemon Exceptions that are encountered when ticking processes only got logged to the daemon log but did not go through the logger of the process itself and therefore the exception message was not in any retrievable through the WorkCalculation node. In the tick_workflow_engine of the daemon, when an exception occurs in the ticking of the process, a log message is sent to the logger of the process which should have the DbLogHandler configured, which will cause the message to end up in the DbLog table. This message will then be accessible through for example 'verdi work report' 10 November 2017, 13:26:04 UTC
96f8999 Merge pull request #898 from szoupanos/fix_for_895 Fix for #895 09 November 2017, 14:03:20 UTC
c2c2a42 Putting back execute_steps() in the manual_tick_all() 09 November 2017, 13:53:20 UTC
fff52c9 Merge branch 'release_v0.10.0rc3' into fix_for_895 09 November 2017, 13:31:06 UTC
666bce8 Merge pull request #904 from sphuber/fix_902_incorrect_stepper_recreation_if_construct Fix recreation of stepper in workchain 'if' logical block 09 November 2017, 13:20:50 UTC
f250f86 Fixes 902 There was a mistake in the logic of the 'if' block of workchains when loading. A position varaiable was being used to keep track of how many times that conditional block was ticked. This was being upped each time but also being used to determine which of possible conditional branches the condition was at e.g.: if(...)( <-- pos 0 ) elif(...) <-- pos 1 ) else( ) Long story short, when the condition was reloaded from a saved state it was possible that pos was larger than the number of conditions (usual just one if there's only an if) and it couldn't resume from where it was when it was saved. 09 November 2017, 13:07:40 UTC
0ec7281 Merge branch 'release_v0.10.0rc3' into fix_for_895 09 November 2017, 12:04:21 UTC
318a6e5 Merge pull request #903 from sphuber/fix_886_retrieve_temporary_list Add retrieve temporary list to CalcInfo datastructure 09 November 2017, 08:38:57 UTC
fa2e5dc Only return a retrieved_temporary_folder FolderData if non-empty If no files were specified in the 'retrieve_temporary_list' of the calculations CalcInfo, we should not pass an empty FolderData to the parse_from_calc call of the Parser. Otherwise the parse_with_retrieved call will always get a retrieved_temporary_folder, even if it is empty. By setting retrieved_temporary_folder to None, unless some files have been specified in the retrieve_temporary_list member we maintain the original behavior of the execmanager calling the parser 09 November 2017, 08:10:02 UTC
105f9d6 mplement parsing of output for Templatereplacer doubler test This means that the parsing of the output file can be moved from the daemon test to the actual parser itself 09 November 2017, 08:09:19 UTC
fde637b Move the Templatereplacer parser to specific test folder Also enhance the parser by explicitly verifying that all the files that are listed in the 'template' input node in the retrieve_temp_list key are present in the temporary folder 09 November 2017, 08:09:19 UTC
3eaab23 Add an integration test for the 'retrieve_temporary_list' The daemon integration test leverages the TemplatereplacerCalculation to launch some trivial calculations. We add the output filepath to the 'retrieve_temporary_files' entry of the template dictionary which will cause the TemplatereplacerParser class to check for an unstored FolderData node containing those files, which should be passed to the parse_with_retrieved function called by the execmanager 09 November 2017, 08:09:19 UTC
c65be7d Add a simple plugin parser for the TemplatereplacerCalculation The parser will not do anything, but if the 'retrieve_temporary_files' key was set in the 'template' input node, then it is expected that those files will be temporarily stored in an unstored FolderData node which should be passed in the 'retrieved' argument of the parse_with_retrieved call of the Parser class. In that case, the arguments are checked for the presence of the unstored FolderData node 09 November 2017, 08:09:19 UTC
f2d066b Clean the TemplatereplacerCalculation file 09 November 2017, 08:09:19 UTC
f05e5c6 Add logging for temporary retrieved files by the execmanager Also deleted the link that was added to the retrieved_temporary_folder which was a left over from testing phase. Since the FolderData is never going to be stored, it also shouldn't have any links 09 November 2017, 08:09:19 UTC
7d573c8 Make the used key for the retrieved_temporary_folder variable 09 November 2017, 08:09:19 UTC
8c8e5ec Cleaning up the documentation of members in CalcInfo datastructure 09 November 2017, 08:09:19 UTC
1b31807 Add documentation for the new CalcInfo.retrieve_temporary_list 09 November 2017, 08:09:19 UTC
fd23743 Implement the concept of retrieve_temporary_list in CalcInfo For some calculation one wants to retrieve certain files for parsing but does not want to store them permanently in the repository because for example they may take up a lot of space. To facilitate for this use case we introduce the concept of the retrieve_temporary_list member in the CalcInfo datastructure. Implementations of a Calculation can opt to provide a list of directives, identical to that of the original retrieve_list member, that will instruct the exectution manager to retrieve those files to a local folder for parsing. The difference is that the files listed in the retrieve_temporary_list member will be stored in a FolderData node that will not be stored in the database and which will only be alive for the duration of the parsing. After parsing is completed the retrieved files are garbage collected. 09 November 2017, 08:09:19 UTC
3f600f0 Some minor cleaning 08 November 2017, 11:18:16 UTC
365fe7a Cleaning up 08 November 2017, 11:12:19 UTC
2db05a8 This should be the fix for the error. 08 November 2017, 10:50:58 UTC
ae1d956 Problem seems to be reproduced 08 November 2017, 10:25:43 UTC
62a5459 Merge branch 'release_v0.10.0rc3' of github.com:aiidateam/aiida_core into fix_for_895 08 November 2017, 10:09:32 UTC
ae80cf6 More progress on reproduction 08 November 2017, 10:05:09 UTC
cd64f72 More proger with the test 07 November 2017, 18:56:34 UTC
36c2a13 Merge pull request #889 from aiidateam/fix_for_old_wf_in_sqla_issue_883 Fix for issue #883 07 November 2017, 18:31:23 UTC
6731502 Merge branch 'release_v0.10.0rc3' into fix_for_old_wf_in_sqla_issue_883 07 November 2017, 17:29:45 UTC
8b2fe88 Trying to reproduce the problem 07 November 2017, 17:29:34 UTC
2dad741 Merge pull request #897 from sphuber/fix_896_abstract_node_sub_class_logger Properly namespace the logger for sub classes of AbstractNode 07 November 2017, 17:29:32 UTC
d98e973 Properly namespace the logger for sub classes of AbstractNode The __metaclass__ of the AbstractNode class does some magic to properly set the logger that is configured for AiiDA for all the sub classes, however it did not properly prepend it with the aiida namespace. This means that for sub classes that live outside of the aiida module, their module path will not match the aiida logger and the configured handlers and adapters will not be applied. As a result, for example, in JobCalculations that live in separate plugins, the DbLogHandler will not work as expected. 07 November 2017, 16:57:22 UTC
cf5389a Merge branch 'release_v0.10.0rc3' into fix_for_old_wf_in_sqla_issue_883 07 November 2017, 14:06:11 UTC
0b6e92e Merge pull request #891 from szoupanos/fix_for_890 Adding proper indexes to db_dbworkflowdata (SQLA) 03 November 2017, 21:31:17 UTC
b3c0041 Making test error message more verbose. Adding the indexes to the models too. 03 November 2017, 15:47:10 UTC
fd3eadf Adding proper indexes 03 November 2017, 10:13:07 UTC
29fb0e8 Merge branch 'release_v0.10.0rc3' into fix_for_old_wf_in_sqla_issue_883 03 November 2017, 09:07:21 UTC
fddd4a0 Cleaning the test 02 November 2017, 19:08:24 UTC
73477ba More work on the test 02 November 2017, 18:52:53 UTC
205fffa Merge pull request #888 from sphuber/fix_318_verdi_work_list Improve verdi work list and verdi work report 02 November 2017, 18:22:54 UTC
6da647d Improve the STATE column for verdi work report Recently another terminal state was added for WorkCalculations to be able to support aborting running WorkCalculations. With that the state picture of a WorkCalculation also became a little bit better defined. A WorkCalculation has three terminal states: * FINISHED * FAILED * ABORTED These states are set by the corresponding attribute to the node and are mutually exclusive. If the WorkCalculation is in neither one of these three terminal states, it can be considered by inference to be RUNNING. A node can also be sealed or unsealed. It can only be sealed until after it has reached one of the three terminal states. If a node is sealed but is not in a terminal state, the state is inconsistent and so we print Unknown. This is to prevent from WorkCalculations during which the code excepted will be incorrectly labeled as RUNNING, which would confuse the user. 02 November 2017, 18:06:21 UTC
6fb0f8f Merge branch 'release_v0.10.0rc3' into fix_318_verdi_work_list 02 November 2017, 16:53:38 UTC
cf3e97e Correcting the workflowdata session problem and adding a test that demonstrates it 02 November 2017, 10:42:56 UTC
8a0b83f Merge pull request #881 from sphuber/fix_880_aiida_path_home_expansion Expand the paths in AIIDA_PATH to expand ~ into absolute path 01 November 2017, 12:36:10 UTC
67afd33 Expand the paths in AIIDA_PATH to expand ~ into absolute path 01 November 2017, 12:00:16 UTC
93e7ea5 Merge pull request #879 from giovannipizzi/fix_direct_scheduler Fix direct scheduler 01 November 2017, 11:46:26 UTC
e80377c Fixing the correspondence of the calculation states in the direct scheduler (most processes are in the S state, they should show as RUNNING; also removed unknown states, and checking only the first character that is sufficient for our purposes. 31 October 2017, 22:35:05 UTC
7da7516 First attempt to avoid that the daemon gets locked while submitting via the direct scheduler. Still to test. 31 October 2017, 21:48:44 UTC
a9d5123 Merge pull request #867 from aiidateam/release_v0.10.0rc2 Merge release v0.10.0rc2 into master 27 October 2017, 13:05:08 UTC
69b331c Merge pull request #866 from sphuber/release_v0.10.0rc2 Update the version to v0.10.0rc2 27 October 2017, 12:44:06 UTC
3d0d05c Update the version to v0.10.0rc2 27 October 2017, 12:32:11 UTC
1192151 Merge pull request #861 from sphuber/release_v0.10.0rc2 Release v0.10.0rc2 27 October 2017, 12:27:33 UTC
c0ec28a Add an integration test for workchains on Travis This integration test will submit 30 instances of a WorkChain to the daemon, which will in turn launch a sub WorkChain that will produce an output. At the end of the test, the produced output is compared to the expected output based on the input to the parent workchain. 27 October 2017, 11:05:44 UTC
26a332e Merge branch 'release_v0.10.0rc2' of github.com:aiidateam/aiida_core into release_v0.10.0rc2 26 October 2017, 14:30:19 UTC
ee85f95 Merge pull request #858 from sphuber/fix_831_make_verdi_daemon_stop_backend_independent Remove the backend dependent code from verdi daemon stop 26 October 2017, 14:17:47 UTC
1e1b13d Remove the backend dependent code from verdi daemon stop The verdi daemon stop command was backend dependent because it contained code that tried to reset the timestamps of the legacy workflow stepper. However, this code is also code in the verdi daemon start command and as such is not necessary. Removing this code from verdi daemon stop means that it can now be called without having to load the database environment. This solves the problem that occurs when people upgrade their AiiDA installation and a database migration is needed, for which they needed to stop their daemon, but the necessity of having to load the database environment, triggered the very same database schema check, resulting in an exception. With this change, people can call verdi daemon stop to stop the daemon before performing a database migration, even when there is a database version mismatch 26 October 2017, 14:09:34 UTC
8284836 Distinguish between a DO_ABORT and ABORTED state for WorkChains To allow a WorkChain to be aborted both from within its outline or externally by a user, we separate the request to abort from the actual abort execution. 26 October 2017, 13:28:44 UTC
f14e95a Move _release_process from on_process_finish to on_process_destroy With the recent addition of the possibility of aborting WorkChains and Workfunctions, in which case the process will go from STOPPED to DESTROYED, skipping the FINISHED state, aborted WorkProcesses would never be released, leaving the corresponding pickles in the running directory, which would be picked up again and failing subsequently. Instead we move this functionality from the on_finish method to the on_destroy and check whether the process has_finished otherwise it is only unpersisted 26 October 2017, 13:25:33 UTC
e1a80ad Put try except around seal() in Process.on_destroy() This change also allows to simply check if the process returns has_finished() as True, which means it is either FINISHED_OK or ABORTED, in which case we want to seal the WorkCalculation node. 26 October 2017, 13:23:39 UTC
6496606 Fix bug in _aborted setter of WorkChain The setter would erroneously always set the value to True ignoring the value passed to the setter, leading to WorkChains being aborted whenever loaded from instance state. Also added a check to prevent an already aborted WorkChain from being "unaborted" which should not be allowed. 26 October 2017, 13:07:26 UTC
d8e5e41 Merge pull request #857 from merkys/fix_tcodexporter Tcodexporter: taking into account scheduler-less calculations 26 October 2017, 12:25:06 UTC
ef50a58 Merge branch 'release_v0.10.0rc2' into fix_tcodexporter 26 October 2017, 12:05:52 UTC
884458a Merge pull request #855 from ltalirz/release_v0.10.0rc2 add hint concerning readthedocs 26 October 2017, 12:02:36 UTC
8c72fb6 Merge branch 'release_v0.10.0rc2' into release_v0.10.0rc2 26 October 2017, 11:33:27 UTC
5ff2722 fix typo 26 October 2017, 11:33:01 UTC
26db5b9 Taking into account that get_scheduler_output() and get_scheduler_error() return None if no files are found. 26 October 2017, 11:32:44 UTC
10ff92e Merge pull request #852 from waychal/issue_777_restapi_user_endpoint Merge fixed issue 777 restapi user endpoint into release_v0.10.0rc2 26 October 2017, 11:01:27 UTC
2e55a8e add hint concerning readthedocs 26 October 2017, 10:33:10 UTC
454bc13 Add two items to CHANGELOG.md courtesy of @ltalirz 26 October 2017, 09:41:16 UTC
85077b2 Merge branch 'release_v0.10.0rc2' into issue_777_restapi_user_endpoint 26 October 2017, 09:27:01 UTC
back to top