ONLYOFFICE MAXA: Mixed Values, Blank References and Typed Examples

• Introduced: The first introduction release and date are not established by the checked sources.
• Availability: Currently documented in ONLYOFFICE’s rolling Spreadsheet Editor help; no final supported-version endpoint is established.
• Scope: Ten scalar cases were executed in Desktop Editors Community 10.0.0.169 on Linux x64. Docs, older desktop builds and mobile clients were not tested.
• Checked: October 14, 2026, against the official MAXA reference, source code and saved native results. Additional arithmetic reconstructions are independently calculated.

=MAXA(value1, [value2], ...)

The first input is required and further inputs are optional. The official function entry lists up to 255 values, including numbers, text and logical values, for finding the largest value. It does not fully explain their conversions. Do not type the optional brackets; these examples use English comma separators.

The reviewed official implementation attempts numeric conversion and compares successfully converted numbers. The helper has separate text, Boolean and empty-value conversions. The native cases below establish specific observable outcomes; the source revision is not a claim about every shipped build.

Create MixedMaximum. Put Input and Value in A5:B5, then enter:

Set B12 to text before entering 2 and verify the stored type. Do not store an apostrophe as part of that string. TRUE/FALSE must be logical values rather than text labels. The saved native fixture’s input types were checked explicitly, including the empty B11 and formula-produced empty text in B14.

Enter the following results in G6:G14. These are the observed outputs for the specified desktop build.

G7 and G12 show that the tested ordinary text did not introduce zero. TRUE and FALSE did contribute 1 and 0. The exact numeric string 2 also participated, but that does not establish a parser for every currency string, locale separator or imported label.

G10 explicitly supplies the genuinely empty B11 as an additional cell reference and gives 0. G14 supplies the result of ="" and gives -3. These inputs look empty but produced different results in the test.

For negative measurements, an added zero can exceed every real reading. Here neither −8 nor −3 is zero, yet adding the empty reference changes the maximum from −3 to zero. Do not interpret that zero as a measured value without checking the referenced cells. This observation concerns the explicit extra cell reference tested here; it does not establish every possible blank-range or array layout.

Create VendorExample separately and enter A1:A5 with no heading inside the range:

In C5 enter =MAXA(A1:A5); expect 45. The vendor illustration and the typed native reproduction agree. The largest numeric input is 45, which also exceeds the converted TRUE value of one. Because 45 is positive, this example alone would not distinguish ignoring text from treating it as zero; the negative-value controls above do.

Preserve the original sheet. On a separate NumericModel sheet, put −8,−3,1 in A2:A4. These are the two numeric measurements and the observed mapping of TRUE. In C2 enter =MAX(A2:A4); expect 1, independently reconstructing G8 without mixed types.

Put numeric 0 in A5 and enter =MAX(A2:A3,A5) in C3; expect 0, reconstructing the FALSE/zero cases. Keep the original logical and status columns outside the measurement model when they mean eligibility or completion rather than a numeric observation.

A returned extreme does not establish completeness, select a unique source record or explain why a reading is unusual. For a real report, record which inputs are accepted and keep the source types alongside their values.

The ten native probes cover one Linux desktop build and these specific scalar inputs. Other text grammars, all-empty ranges, arrays, error propagation and Docs/mobile behavior remain outside the tested scope. Recheck them before relying on identical behavior after migration.

MAX supplies the plain numeric comparison used in the reconstructed model. MINA selects the opposite mixed-input extreme. AVERAGEA has a different referenced-text rule. COUNT can help inspect numeric observations, but no single aggregate replaces a stored-type review.