Contact

Collabora Office TOROW: Row Vectors and Filtering Order

Lookup Intermediate Collabora Office Calc

Flatten ranges into a row while keeping scan order, empty text, zero and error filtering distinct in bounded source-checked examples. Scoped to the new Linux desktop source build.

Verification status: See the evidence and test scope described in this guide. How formula examples are checked

TOROW flattens a rectangular input into a single row. The scanning direction controls the order of values; the Ignore argument controls which cells survive. These are separate choices: changing the output orientation alone does not change the source scan order.

Syntax and defaults

=TOROW(Array [; Ignore [; By_column]])

Array is a range or matrix. Ignore defaults to 0, and By_column defaults to FALSE, meaning scan each source row from left to right before moving down. TRUE scans each column from top to bottom before moving right.

Ignore Source cells removed
0 None
1 Genuine empty cells
2 Error-valued cells
3 Both genuine empty cells and errors

The source checks an empty-cell type, not simply whether a cell looks blank. A formula returning empty text and a numeric zero are not interchangeable with a genuinely empty cell. The examples keep these categories explicit.

Step by step: enter a typed two-by-three input

Set up A2:C3 as follows. Leave B2 genuinely empty; do not type a space or a label there.

Cell Enter Meaning
A2 10 Number
B2 nothing Genuine empty cell
C2 0 Numeric zero
A3 ="" Empty text result
B3 =1/0 Deliberately error-valued input
C3 20 Number

This is an instructional setup. Neither these input formulas nor TOROW were executed in an application for this guide.

For =TOROW(A2:C3), the source-derived sequence is 10, empty cell, 0, empty text, division-by-zero error, 20, arranged as one row. With no filtering, the six source positions become six output positions. The empty-cell marker in this explanation is not a text string to enter into the worksheet.

The evaluator’s copy helper preserves a genuine empty matrix cell internally. Its eventual display in the new desktop was not tested; no promise that it will appear as a particular blank, zero or formatted string is made.

Filter blanks and errors separately

=TOROW(A2:C3;1) removes only B2 and retains five positions: 10, 0, empty text, division-by-zero error, 20.

=TOROW(A2:C3;2) removes only the error and retains 10, empty cell, 0, empty text, 20. Choosing 2 is not a general “remove empty-looking values” operation.

=TOROW(A2:C3;3) removes both the genuine blank and the error. The resulting four values are 10, 0, empty text, 20. In this control, numeric zero remains real data and empty text remains a text value.

Filtering compacts the sequence. The later entries move to fill removed positions, so their new output indices no longer match the original flattened positions. Do not use the output index as an unchanged source-cell address.

Change scanning direction explicitly

=TOROW(A2:C3;3;TRUE)

The source visits A2, A3, B2, B3, C2 and C3. After removing the blank and error, the sequence is 10, empty text, 0, 20. Compare that with row scanning’s 10, 0, empty text, 20: the same four surviving values occur in a different order.

The form =TOROW(A2:C3;;TRUE) leaves an explicit placeholder for Ignore, which the checked evaluator defaults to 0. It retains all six entries in column-scan order: 10, empty text, empty cell, division-by-zero error, 0, 20. Keep the separator so TRUE remains the third argument.

Advanced: preserve the meaning of the flattened data

A single-row layout can provide a horizontal comparison strip. If source rows represent categories and columns represent periods, row scanning places each category’s periods together; column scanning groups categories within a period. The function does not infer which arrangement matches the intended chart or report.

For R×C input with nothing ignored, the result has R*C columns. Ignore=3 reduces the typed example to four columns. Pairing that compacted row with six unfiltered headers would mislabel values. Filter or rebuild the headers with the same explicit rule rather than assuming position has retained its original meaning.

Empty results, errors and array workflow

With H2:I3 all genuinely empty, =TOROW(H2:I3;1) removes every cell. The source returns internal NotAvailable, rather than constructing an array of length zero. Ignore=4 reaches internal IllegalParameter because the supported converted values are 0 through 3. These are evaluator identities, not observed localized error messages.

The help describes array entry and a Ctrl+Shift+Enter hint. This guide establishes source matrix ordering and shape only, not that shortcut, automatic spill or resizing in the new Qt interface. With N retained entries, the result shape is 1 row by N columns. TOCOL uses the other output orientation; it does not make an arbitrary two-dimensional transpose equivalent to flattening.

Product scope and verification

This guide covers the new Collabora Office desktop for Linux, source build coda-26.04.4.3-1, commit 81339d176148d0d5c81ff05d82540f58f8aaa117, mapped from the vendor’s Linux Flathub package. The 26.04 TOROW help and exact evaluator source were checked on October 18, 2026. The desktop announcement and package manifest explain the narrow product mapping; the manifest is a mutable source publication, not proof of an installed binary.

No Collabora runtime or other spreadsheet engine was executed. The matrices are independent/source-derived expectations, and the typed input setup is instructional. No native screenshots or interface tests are represented. Classic, Online, CODE, mobile, Windows and macOS parity is not established.

The original Collabora introduction release is unknown from the checked evidence. The help identifies COM.MICROSOFT.TOROW and places the function outside OpenFormula 1.3; that compatibility limitation is not a retirement or deprecation notice. Import/export and recalculation in other products were not tested.