I am the first to point out, that I (and likely many of readers on the Cove) live in the USA and our comments on GD&T are based on American drafting standard ASTM Y14.5. At one point in my career I had occasion to work with the DIN standard. I don't know what standard is commonly used in Poland. The latest version of Y14.5 is 2018, but there have been differences from earlier versions (2009 and 1994) so correct interpretation may depend on which GD&T standard applies to the particular engineering drawing you are asking about.

@Magdalena is living in Poland (welcome!) so I will use the comma as decimal separator, as is common practice in European countries.

my part has a bracket which size is 30mm. Its still says Surface Profile 0.4, but Flatness is 0.5? so full plane of brkt must be 0.5?

The information in the first posting by

@Magdalena was Surface Profile 4, so I am guessing this later posting of 0,4 was a mis-type.

Flatness is not the same as Surface Profile, as explained in this

link. In simple terms, a flatness callout limits deviations from a virtual ideal plane (the plane which best fits the physical surface). A Surface Profile callout is used relative to one or more datums, and controls shape, location and orientation in 3-dimension space, while Flatness controls only planarity (to itself). I suppose a Surface Profile callout with no datum references becomes more similar to a Flatness callout, in theory, but there are other differences as noted in the earlier link.

I don't agree with

@JanKees example: in a product of 400 sq mm, there are actually an infinite number of sections of 100 sq mm (one starting at the top edge, one stating 1 mm from the edge, one starting 2 mm from the edge, and so on). All

**possible **sections of 100 sq mm must conform to the 0,5/100 requirement for the product to be in compliance. That would be my interpretation. In practice, the entire surface would be sampled by a CMM, or perhaps the greatest distortion from a flat plane could be identified (if there was a reliable way of doing this, maybe visually) and then 100 sq mm samplings containing the greatest distortion would be compared to the 0,5 mm limit.