If I need more proof of the lack of understanding of the problem from the general user population, here is a quote form today’s Linkedin:
"The problem is that without structure (schema) one needs to program in a procedural language.” Author will remain unknown…. Dana > On Jun 1, 2015, at 11:40 AM, daniela florescu <[email protected]> wrote: > > >> On Jun 1, 2015, at 10:51 AM, Michael Kay <[email protected]> wrote: >> >>> >>> Using SQL to process JSON is simply NOT POSSIBLE. >>> >> >> Proof? >> >> My Ph.D. thesis in 1975 showed that you can use SQL to query hierarchic >> data. With the emphasis on “query”. The problems start with update. So long >> as you restrict yourself to query, you can map almost any data model onto >> any other. > > Michael, > > you did not read my email properly. > > I did NOT say that if you map JSON to relations (one way or another), you > cannot use SQL on the result of the mapping. > > I said: you cannot apply SQL DIRECTLY on the JSON structures, because it’s > semantics is not designed for that. > > Com’on Michael. You worked with me for 16 years now. Do I look like I’m so > naive and I don’t get things ? > > So, please. I understand that EVERYTHING (C++ objects, XML elements, COBOL > structures, > assembly processing instructions, anything) can be mapped with enough brute > force into relations, and that, without > loss of any information. > > So thank you for this notice, but it does not relate to my email. > > (BTW, I proved the same thing as you 2O years later…and this was 20 years > ago…didn’t know about your PhD sorry…. > http://homes.cs.washington.edu/~suciu/PAPERS/florescu-kossman.pdf > Apparently this is something that CS has to get it out of their system every > 20 years …the time is up again, I guess…...) > > > The problem is that Drill (like everybody else in that bunch) apply SQL > DIRECTLY on JSON structures, in the absence of any intermediary mapping > into relations, which is again, I repeat: A NON-SENSE. > > > And BTW Michael, if SQL would be good for querying hierarchical data, why in > the world did we waste 16 years building XQuery !??? > > > Just curious > Dana > > > > > > >> >> Though the fact that you can do it doesn’t mean that it’s a good idea, of >> course. In those days the reason for doing it was that SQL was the only >> declarative query language in town, and that’s no longer the case. >> >> Michael Kay >> Saxonica >> >
_______________________________________________ [email protected] http://x-query.com/mailman/listinfo/talk
