Contact

Calligra Sheets Β· Data management Β· Intermediate

Filter Records Without Losing the Full Total in Calligra Sheets

Apply and clear two Auto-Filter fields on a disposable table, verify whole-record visibility and keep filtered views separate from the full-data total.

Pinned source reviewed; native verification pending Β· Editorial standards

Filter a six-record work queue by region and stage, then restore every record. The important distinction is between which records are visible and which records a formula totals. Removing filter buttons is also different from clearing the active criteria.

Scope and safe preparation

This is a source-qualified workflow for English Linux/KDE-oriented Calligra Sheets, based on official source commit 72123cc643f6625351139e7df294544255a1056f. It is not a native application test. The older KDE handbook identifies itself as revision 2.7, March 11, 2013; its stable_kf6 address is not proof of a modern interface. The particular filtering controls below are established by pinned menu, action and popup source, not inferred from the historical Sort guide.

No new native GUI run, screenshot, filtered-view persistence or import/export round trip was performed. Other builds and translations need separate checks. If the named controls or expected record identities differ, stop and retain the baseline.

Use a new empty sheet. Filtering changes whole-row visibility, including unrelated cells alongside the table, so do not experiment next to another report. Put audit cells on row 10 or later, below all filtered records. Use literal labels and numeric amounts, with no merged cells, hidden rows, blank records or pre-existing filters.

Enter six complete records

Enter these cells individually in A1:D7, including the header in row 1:

Row A: Record B: Region C: Stage D: Units
2 R01 North Open 10
3 R02 South Open 20
4 R03 North Done 5
5 R04 South Done 15
6 R05 North Open 7
7 R06 South Open 3

In F10 enter Full-data units, and in G10 enter =SUM(D2:D7). Expect 60. Count the six record IDs and check every ID/region/stage/units association against the table. G10 is a total of the full bounded data range, not a visible-record total.

Save filter-queue-base.ods, then Save As filter-queue-working.ods. Keep the baseline unchanged. Filtering is a view operation, not a reason to delete the records you do not want to see.

Add buttons to the exact table

  1. Select A1:D7, including the header and all six records, but excluding the audit cells.
  2. Choose Data β†’ Filter β†’ Auto-Filter once. The pinned action creates a database range with filter buttons. Its default orientation is vertical and its first row is treated as the header.
  3. Check that filter arrows appear in the header row. If no arrows appear, stop and compare the installed version rather than repeatedly toggling the command.

Use the explicit rectangle rather than a single selected cell. A single-cell invocation can reuse an existing database or expand to a surrounding range; that inferred selection is not the bounded fixture described here.

Show the North records

  1. Click the filter arrow in B1, Region.
  2. In its popup, clear All to clear the available selections, then check North only. Leave South and Empty off. The popup also has Non-empty; do not select it for a North-only filter.
  3. Click outside the popup to close it. In the reviewed implementation, changed selections apply when the popup closes; there is no Apply button to wait for. Do not assume closing or pressing Escape is a guaranteed cancellation route.
  4. Check the visible IDs and their complete rows:
    • R01 β€” North β€” Open β€” 10
    • R03 β€” North β€” Done β€” 5
    • R05 β€” North β€” Open β€” 7

The independent visible-record sum is 22. G10 should still total the full range at 60. The original row numbers should remain 2, 4 and 6; filtering does not renumber or sort the records.

If a record’s units appear beside the wrong ID, stop: a filter must not be accepted merely because its numeric total looks plausible. Recover from the preserved baseline, not by guessing which rows belong together.

Combine a second field

Click the arrow in C1, Stage. Clear All, check Open only and close the popup. The source combines conditions across fields: North and Open should leave these two records visible:

  • R01 β€” North β€” Open β€” 10
  • R05 β€” North β€” Open β€” 7

The independent visible-record sum is 17; the full-data G10 remains 60. R03 should disappear because it is Done, even though it is North. R02 and R06 should remain out of view because they are South, even though they are Open.

Do not use the unchanged 60 as the filtered workload. It intentionally includes all six numeric cells. For this small fixture, independently add the units of the explicitly verified visible IDs. A larger production report needs a separately reviewed aggregation workflow rather than a manually counted sample presented as scalable automation.

Do not assume SUBTOTAL solves the filtering distinction here. The existing SUBTOTAL reference has a separate manual-hide test. In the pinned source, filtering sets a filtered-row flag; manually hidden rows use another flag. The greater-than-100 SUBTOTAL branch calls the hidden-row check, not a combined hidden-or-filtered check. This guide therefore makes no filter-aware SUBTOTAL guarantee and does not promote that historical manual-hide evidence into a filtered-total result.

Clear criteria one field at a time

  1. Open the Region popup and select All, then close it. This clears Region’s criteria only. Stage is still Open.
  2. Verify that the visible IDs are now R01, R02, R05 and R06, with units 10, 20, 7 and 3. Their independent sum is 40, while G10 remains 60. Seeing four rather than six records is expected at this intermediate step.
  3. Open the Stage popup and select All, then close it.
  4. Verify that all six IDs are visible again in their original row positions, with every value unchanged and G10 equal to 60. Inspect both popups if any record remains out of view.

Selecting All removes the conditions for the field whose popup you opened; it is not a workbook-wide β€œclear every filter” command. The Region reset alone must not be recorded as a complete reset.

Hide buttons only after clearing the view

After restoring all six records, select A1:D7 again and choose Data β†’ Filter β†’ Auto-Filter once if you want to hide the arrows. For an exact matching range, the reviewed action toggles button visibility. It does not clear field conditions or remove filtered-row state.

If you accidentally hid the arrows while records were filtered, do not treat the table as unfiltered. Select the same A1:D7 rectangle and toggle the arrows back on, then use All in each affected field as above. If the original range is uncertain, recover a separate working copy from filter-queue-base.ods rather than creating overlapping filters.

Save the fully restored table as filter-queue-final.ods. Close and reopen a copy; verify all six records, their associations and total 60. Reopening a filtered file and retaining/reconstructing its criteria is not established by this exercise. A successful save alone would not prove that behavior.

Add a record from a clean source

Open a copy of the baseline, add R07, North, Open, 5 in A8:D8 and save under a new name. Change G10 to =SUM(D2:D8); expect 65. Add Auto-Filter to A1:D8, not the earlier seven-row range.

Repeat the same choices. North alone should show R01, R03, R05 and R07, totaling 27 by independent addition. North and Open should show R01, R05 and R07, totaling 22. Clearing Region alone leaves five Open records totaling 45; clearing Stage too restores all seven records and the full-data total 65.

This deliberately rebuilds from an unfiltered source. It does not assume automatic filter-range growth, live refresh when a filtered value changes, or safe reuse of an overlapping existing database range.

Primary sources and remaining limits

The fixture’s record sets and arithmetic were independently checked outside Calligra. Case variants, blank-field behavior, manually hidden-row interactions, custom-filter dialogs, persistence and other applications remain unverified. For actual record reordering rather than visibility, use sorting with headers.