Avoid parallel table.close() on the same table #25
Conversation
|
Time to go back to this PR? |
|
Tables are shown as expected on py3-ci and merge-ci. I tried (on merge-ci) both the orphaned dataset indicated in the PR header as well as the idr0021 project table. On py3-ci, I have tried only the idr0021 project table. I found the "waterfall" feature of the developer tools in chrome (at least temporarily) but did not see any of the green bars as shown in the header of this PR - maybe I need ot readjust something. ? |
|
@pwalczysko : i.e. you need to help to finish this review? |
|
@will-moore helped me to find the "waterfall" feature in the browser. Again, confirmed that I am not getting the errors in the UI and also, my screenshots (see below) look as expected in the sense of the two calls for image and Image waiting for one another as expected (i.e. the image call waits for the Image call to finish before being sent off). This was verified by reading the tooltips on the lines in the "waterfall" table. Tested both table on a Project as well as table on a Dataset.
Ready to merge fmpov |
For tables on Projects or Datasets, we try 2 queries
image=1(lowercase) andImage=1on the same Table at the same time, which leads to errors. See #6This changes the webclient right panel to make each pair of queries run in series instead of parallel.
Screenshot shows calls in Chrome dev-tools before this PR, where we try to load table data on a parent Screen, Plate, Project (x2) and Dataset (x2) all at once.
With this PR, we wait for the first queries on Project and Dataset to return, and if they have no data then we make another call with the alternative query:
To test, check that table data is shown as expected with no error dialogs.
This can be tested at https://merge-ci.openmicroscopy.org/web/webclient/?show=dataset-4919 (user-3), which was the sample data originally used to replicate the error.
NB: In this data, the first call to get table data from Dataset is successful, so a second call is never made.