Dyalog stores code points directly using 1-, 2-, or 4-byte unsigned integers. The type for a given array has to fit all the characters, and we try to choose the smallest possible. I'm not sure how good our surrogate pair handling is, but I think they are supposed to be combined into single characters on input.
Marshall On Thu, Sep 19, 2019 at 12:18:13AM +0100, Ian Clark wrote: > Well done, Bob. > > I've read the "differences between revisions" and that's a mean task you've > completed. > > I have to confess I find the new stuff totally baffling. I wrote the > original article 2 years ago and I still have the bruises on my forehead :) > I was ignorant of how J901 supports the newer code pages until I read it on > this thread. > > Some helpful(?) questions: > ++ How does Dyalog APL do it? > ++ How does Swift 5.1 do it? > ++ How does Python 3.7 do it? > ++ How does Javascript do it? > …All are languages with serious pretensions to manipulating text containing > UCPs. Maybe over 90% of application code being written in these languages > does just that, and mostly on webpages. The writer of the Swift manuals > published by iBooks delights in showing emojis between quotes in code > samples. Smart stuff – but only a GUI coder or indie publisher would know > it. > > In my day-to-day programming I have little or no use for any greater > precision than utf-8 and wide characters (…are we still calling them that? > – how about mega-wide and giga-wide for the new precisions?) Just about the > only use I'd have for the newer UCPs is to embed them in a PDF document via > copy-paste. Nowadays that's more likely to be a layman's review blog than a > learned paper. In which case I'd be at the mercy of my WP vendor to get it > right when coding the copy/paste. > > On past form, the omens are not good. From 1999 to the present day, as an > indie publisher of books with fancy fonts, I watched Microsoft and Adobe > completely foul-up the introduction of utf-8 to their products, notably > export to PDF. Assuming it won't take them another 20 years to migrate to > utf-32, I guess I can look forward to running sequential machines on emojis > in my care home. > > Ian > > On Wed, 18 Sep 2019 at 20:45, 'robert therriault' via Programming < > [email protected]> wrote: > > > Hi Henry, Bill and Ian > > > > I have edited the wiki for the UCP page. > > > > The synopsis is that I included some information on how literals and utf-8 > > are related and a section on surrogate pairs. I hope I got most of this > > right, but if I didn't please make the necessary changes and/or correct me. > > > > Ian, I hope that I was able to retain the spirit of what you established > > with your excellent foundation. > > > > https://code.jsoftware.com/wiki/Vocabulary/UnicodeCodePoint > > > > Cheers, bob > > > > > On Sep 13, 2019, at 10:59 AM, Henry Rich <[email protected]> wrote: > > > > > > Detail is great, but put it towards the end of the page if possible. > > > > ---------------------------------------------------------------------- > > For information about J forums see http://www.jsoftware.com/forums.htm > > > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
