Talk:Hexadecimal floating point

This isn't really a disambig page or a set-index-article

@Matthiaspaul: "Hexadecimal floating point" is a concept implemented by multiple vendors, of which IBM's implementation is by far the most notable. We have an article on IBM's implementation specifically, and this page redirected to that. What you seem to be doing here is trying to create a separate article on the general concept. It isn't really a disambiguation, because we don't have articles on those other implementations, and we probably never will (none of the non-IBM implementations of HFP are notable enough to deserve a distinct article about their HFP implementation specifically.) It isn't really a set-index article – set-index articles are for things of same type that have similar names, but "Hexadecimal floating-point arithmetic in the X computer" isn't even a name. Maybe, a better way to achieve what you are trying to do, is rename the IBM article to just "Hexadecimal floating point", and split it up into three main sections: (1) Introduce the concept in the abstract (2) Provide a summary of all historical implementations of this concept (3) Provide extensive details on IBM's implementation in S/360. So (2) would basically be what you've created here, (3) would be most of the current IBM-specific article, and (1) would be some of the introductory material from the IBM-specific article. SJK (talk) 10:13, 19 August 2019 (UTC)

Hi, your description is quite close to what I'm trying to achieve. As this is a very broad theme the goal is to have an article about floating point in general and individual articles on the different bases and sizes. They should unfold the whole story including historical implementations, their shortcoming and special tricks, up to what we use today (if applicable) and why. All aspects of IEEE 754 should be covered as well, but not as if there would not have been other implementations before (or even now) - some of the current wording could be read as if there would be nothing else, and since we are an encyclopedia, I think, we have a duty to give our readers more background and put everything in context.

Regarding the hexadecimal floating point list article, yes, I envision this to become a generic article on hexadecimal floating point over the years. For some implementations there is little information available so that their hex floating point formats could be covered in their individual articles about the machines, whereas in the case of the IBM 360 series, hex floating point stuff is so prominent that it deserves a separate article (as now).

Like you I was also thinking about renaming the IBM hex floating point article and adding more generic stuff there as well, but felt that it might confuse or even upset some readers who want to learn about the IBM implementation specifically. Might be similar to putting too many IEEE 754 specifica into the generic floating point article (although they should definitely be mentioned in the IEEE 754 articles). Unsure...

So, my idea for now was to at least prepare the basis and infrastructure at hexadecimal floating point, and see what happens. Continuing to collect more infos on the individual implementations - and once we have collected enough material, probably not before a couple of years, to grow this into a full-blown article.

--Matthiaspaul (talk) 10:53, 19 August 2019 (UTC)

How about, instead of calling it a "disambiguation" or a "set index article", we simply make it a stub? Leave the IBM hex floating point article as separate, but maybe copy a little bit of introductory material to here. Keep your list of other platforms having hex floating point. Stick a {{stub}} template at the end. SJK (talk) 11:46, 19 August 2019 (UTC)