The run list
The Runs list and its filters, the row anatomy, paging and refreshing, the import control, and deleting one run or a whole sweep.
The left half of the browser is headed Runs, with two controls in its head: Import
(a .glab file recorded elsewhere, Importing runs) and Refresh.
Under it the filter strip, the bulk bar when runs are ticked, one page of 50 runs, most
recent start first, and the pager ("1–50 of 3214"), drawn only while at least one run
matches. There is no other sort order. Selecting a row opens the run on the right and marks
the row.
Refresh reloads the page and re-reads the Unit and Responsible choices, so a unit configured or a user created since the panel opened becomes filterable without leaving the browser. The list does not refresh on its own: a run recorded while you read another one appears at the next Refresh, filter change or page step.
The row
| Element | Meaning |
|---|---|
| Run name | The name the run was recorded under: "Procedure · Unit" for a procedure run, "Snapshot · Datalog" or "Evaluate · Evaluation" for a single command. |
| Automation badge | The run was admitted by the unit's automation interface, not a person. |
| Imported badge | The run arrived as a .glab file from another installation. |
| Green dot | The run is Running. |
| OK, NOK or Indeterminate badge | The verdict. A run with none carries no verdict, or has not finished. |
| Sub-line | The lifecycle state, then Interlock when the unit's external stop signal ended the run, then who started it, then the start instant in the station's format. A narrow column trims the responsible's name; the whole line is in the row's tooltip. |
| Checkbox | Ticks the run for bulk deletion ("Select for deletion"). Drawn only for a role with Manage histories. Greyed for a run that has not finished ("A run that has not finished cannot be deleted.") and on a read-only station. |
Filters
Every filter is a column of the run index, the search box included: the index keeps each run's metadata values beside its name. So the strip resolves as one indexed query: no run file is opened to decide a match, and an archive of years never loads whole. Changing a filter returns to the first page.
| Filter | Choices | Notes |
|---|---|---|
| Filter runs by name or metadata value | Free text | A case-insensitive part of the run name or of any metadata value the run was started with (a serial number, a lot), applied as you type. |
| Verdict | All verdicts, OK, NOK, Indeterminate, No verdict | No verdict lists runs that carry none, finished or not. |
| State | All states, Running, Held, Completed, Aborted, Failed | Running also lists a run still writing its file (Saving), which has not stopped recording as far as you are concerned. |
| Unit | All units, then one entry per unit as "Model · Unit" | Where the run happened, in model order then unit order. An imported run belongs to no unit and lists only under All units. |
| Responsible | All responsibles, Station, Automation, Agent, then every local user by name | Who started the run. A person is matched by the identity frozen into the run, not by the name: a renamed operator keeps their runs, and a deleted one leaves the choices while their runs still name them. |
| From / To | Calendar days | The earliest and latest run start to list. Both are station-local days; the To day is included whole. |
When no run matches, the list draws neither rows nor a pager, only one line in their place. While any filter is set, the search box included, the line reads "No runs match the filters.", even on a station that has recorded nothing; with every filter cleared, only a station with no run at all matches nothing, and the line reads "No runs recorded yet." Until the first page of the index has been read, the line reads "Loading the runs…"; when the index cannot be read, it says so, with the reason.
Runs recorded by a version whose index did not keep metadata values get them once, in the background, from the first start after the update. The station reads one run file at a time and waits while a run is recording or the runtime is stopped, so nothing is held up. Until it has finished, a search finds those runs by name only and says so under the box: "The search does not cover the metadata of older runs yet: they are still being indexed, and until then they are found by name only." A run file that cannot be read is skipped, and that run is found by name only until Rebuild run-history index reads it again.
Paging
Previous page and Next page walk the match a page at a time; the ticked set and the open run survive a page step. Deleting the tail of the archive moves the position back to the last page that still exists.
Deleting runs
Deleting a run takes Manage histories. A role without that permission is drawn no tick, no bulk bar and no Delete, so the gesture is absent rather than greyed.
Delete, the danger verb at the end of the record's strip, removes the open run and its file after "Delete 'name'? This removes the run and its file. Documents already published for it stay in the reports folder. A run that has not finished cannot be deleted." It is greyed on a read-only station and for a run that is Running, Held or Saving; the service refuses a run that still owns its source locks with the same rule. The outcome lands in the action feed ("Deleted · name").
For a sweep, tick runs in the list: a bulk bar appears with "N selected", Page (tick every deletable run on this page), Clear (untick every run) and Delete (greyed on a read-only station and while a sweep runs). The ticked set is kept by run identity and survives paging and refreshing, so a date-filtered sweep can be assembled page by page before one confirmation ("Permanently delete N recorded runs and their files? Documents already published for them stay in the reports folder. This cannot be undone.") deletes it. Each run is deleted on its own, index row and file together, so a run the runtime still owns is refused individually and the feed reports "Deleted X of Y runs · Z refused: reason" instead of failing the whole sweep. Deleting a run also removes its contribution to the unit's productivity. It releases the run's export file name too, unless a document published for that run is still in the reports folder: the documents stay there by design and carry that name, so it remains taken while they do, and no later run can publish over one of them.
What the list does not do
It does not sort by anything but the start instant, does not search a run's comments or recorded values (only its name and its metadata values) and does not open two runs at once; a second browser tab is a second browser with its own selection and filters.