SC22/WG15 N464 From: NNI To: SC22/WG15 May 4, 1994 Subject: Action item responses Date: May 1994 Responses from NNI on action items 9305-07 SGFS Liaison Status: done. P1003.0 D16 was circulated as SGFS/N1030 P1003.0 D16.1 was circulated as SGFS/N1078 P1003.18 D10 was circulated as SGFS/N1107 9310-14 Member bodies NNI has no further comments on N375 (the list of old action items) and recommends to close this action item. 9310-20 RGCPA NNI offers the comments, as given in Annex A of this document, on the latest draft of the POSIX National Profile and National Locale. 9310-24 Member bodies As NNI will host a WG15 meeting in April 1995, NNI cannot offer to host a WG15 meeting in 1996. 9310-26 Member bodies There are no Dutch national profiles to offer to WG15. 9310-27 Member bodies The documents, mentioned in this action item are old, and overtaken by events. We recommend to close this action item. Our comments on the relationship between P1003.0 (PDTR 14252) and TR10000 are in Annex B of this document. 9310-37 Member bodies NNI has no participants to offer for the 9945-1 and 9945-2 interpretations process. 9310-40 Member bodies We have no comment on 2003.2. ------------------------------------------------------------------------ Annex A Comments on the POSIX National Profile and National Locale document Issue 1: the relationship with TR10000. We have the feeling that the response from SGFS to WG15 on the issue of National Profiles was misunderstood. SGFS did not see a need for a special class for National Profiles. However, this does not mean that National Profiles are not Profiles in the TR10000 sense. This means that National Profiles, like all other SGFS Profiles, should follow TR10000, and especially TR10000-1. When WG15 considers it necessary to have special requirements for National Profiles (and these can be very legitimate), then the National Profile TR should first include (by reference) TR10000-1 (and not just the subset of section 6.3.1), and then specify the additional requirements. The current draft of this TR simply repeats large portions of the text of TR10000-1 which is not a good idea: this suggests that requirements on National Profiles are essential different from those on `normal' Profiles. As this is not the case (or should not be the case), there should be only one document describing the `common' requirements. Issue 2: what is defined by a National Profile? We have the feeling that what is needed is a Profile that describes an incomplete product: it refers to the base standards, with their options (especially the I18N option), and the interface to the National Locale. The Profile is made to a complete specification by National Member Bodies by specifying the National Locale (thereby making it a `real' National, i.e. US, Japanese or Danish, Profile). The requirement for an `incomplete' ISP is not new: the Common Upper Layer Requirement Profile (CULR, ISP 11188-1) is also such a profile. So, the `International' part of a National Profile can very well be (and should be!) an ISP. Having said this, the question arises: what is a National Profile, as described in this draft? Is it the International subset profile, or is it the localized (nationalized) profile? Some minor points: - the current thinking goes towards profiles that refer to other profiles. I think that it would be (more?) useful to have the National Profiles refer to the POSIX Profile (like .18), let .18 define the relationship between 9945-1 and 9945-2, and have the National Profile define the `overall' national requirements. - has any consideration been given to application of the National Profile concept to areas outside POSIX? ------------------------------------------------------------------------ Annex B Comments on the relationship between PDTR 14252 (P1003.0) and TR10000 NNI supports the classification of OSE profiles, as currently proposed for TR10000 part 3 (see OIW OSE-TC/94.051 and EWOS/EG-OSE/94/34), based on the four types of interfaces (API, ISI, CSI and HCI). There is no problem in using the names of the interfaces, defined in PDTR 14252, in TR10000: the POSIX OSE, as described in PDTR 14252, describes a restricted view on OSE with as main purpose portability and interoperability. The view on OSE, as described in TR10000 part 3, is wider, and also includes the identification of interfaces that allow for products to be constructed in an incremental fashion. Furthermore, TR10000 is described (in the proposals) in an architecture neutral way. It therefore makes perfectly sense to generalize the definitions and concepts from PDTR 14252 in TR10000, but still to use the same names (API, ISI, CSI and HCI). To avoid confusion, NNI will propose to SGFS to select the 2nd option from the EWOS proposal on TR10000-3 (the inclusion of the generalized definitions of the four types of interfaces in TR10000). On the issue of PIIs, it is our opinion that software PIIs are just a special case of APIs, and are defined in PDTR 14252 in order to be able to exclude them from the POSIX OSE Reference Model. This exclusion is valid for PDTR 14252, but as a consequence of the generalization of the four types of interfaces in TR10000, this special treatment is not necessary (and even not wanted) in TR10000. NNI believes that all these (types of) APIs are within the scope of TR10000 part 3, and that no special treatment of PIIs is necessary in TR10000. Although it can be envisaged that PIIs might play a role in future POSIX OSE profiles, NNI would oppose any proposal to include PIIs in the POSIX RM at this time: it is most important that PDTR 14252 gets stabilized and approved, so that experience with its application can be gained. If necessary, this topic might be listed for a future revision.