IFPUG – NESMA differences

[Han Suelmann]

I think it would be a good idea to re-examine differences between IFPUG 4 and NESMA 2.2 FPA. The latest I have heard on this subject is the document “FPA volgens NESMA en IFPUG; de actuele stand van zaken” (“FPA according to NESMA and IFPUG; current status”), dated August 1, 2011. The document does contain a list of differences, but I think that the list is incomplete. Specifically, I see two things that seem to be missing.

First, there is a difference in the way that batch processes are treated. NESMA 2.2 explicitly tells us to count them as EO’s; IFPUG 4 explicitly tells us not to. This leads to NESMA counts that are higher than IFPUG counts, the difference depending on the nature of the software.

Second, there was a difference in the way in which reports in general are treated. NESMA 2.2 this criteria for distinguishing, under certain circumstances, several EO’s for one report. I cannot find the same thing in IFPUG 4. I have not examined in detail to what differences in the final count of this might lead.

I do not really know whether there are other significant differences. It might be a good idea to check this. For analysis purposes, it would probably be useful to have a numerical indication of the resulting differences: how much higher or lower is a NESMA count on average, compared with IFPUG 4, in daily practice?

[Theo Prins]

I don’t agree with the the first statement of Han. In both counting guide line documents I can’t find any foundation for this. Han writes ” NESMA 2.2 explicitly tells us to count them (being batch processes) as EO’s; IFPUG 4.x explicitly tells us not to”, but this can not be found (by me) in the guide line documents.

Further I wish to add a third statement. In the document “FPA according to NESMA and IFPUG; current status” there are some remarkt about code data (IFPUG 4.x) in relation to FPA-table (Nesma 2.1). From the point of view of IFPUG code data is an implementation of non-functional user requirements. As a consequence, code data does not influence the functional size of the software product. According to the document “FPA according to NESMA and IFPUG” (1) in the case of primary data, both NESMA and IFPUG follow the same counting guidelines and (2) code data tables are identical to FPA-tables. Both claims are not entirely true. IFPUG distinguishes three categories of data entities: Business Data, Reference Data and Code Data. The first two categories of entities (primary data) are usually identified to satisfy the Functional User Requirements, and as such these entities will be investigated for counting as logical files. In a Nesma setting Reference Data tables are identified as EIF/ILF’s or as FPA-tables (EIF/ILF), depending on the characteristics. (E.g. an entity type contains boundary values, algorithms, and minimum or maximum values, provided that the key is single is an FPA-table (Nesma, type 4) and a Reference Data table (IFPUG). Therefore it is possible that in an IFPUG-analysis Reference Data tables are identiefied as distinguished ILF’s and in a Nesma-analysis as one FPA-table.
On the other hand, in a Nesma-analysis, code data tables are not by definition an occurence of an FPA-table. Sometimes these tables are system tables and for that reason ‘ignored’ in a Nesma-analysis.

[Han Suelmann]

The NESMA 2.2 rule is in 8.2.p.
IFPUG 4.3.1 discusses the issue in Part 4, in the example starting on page 2-4.

I agree, by the way, with the third statement you added.

[Theo Prins]

The references to the guidelines make the first statement clear. The point is not the way that batch processes are treated, but the manner in which run reports of batch processes should be counted. I agree there is definitely a difference in approach between Nesma and IFPUG on this issue: 8.2.p (2.2) / 9.2.p (2.1) indicates to identify one or more EO’s, while IFPUG (4.3, P4, 2-41 a.n.) states that the batch report is not an elementary process, which should not be counted.

[Author]

Posts