On 2016/05/13 5:24 AM, dandl wrote: >>> Richard Hipp seems to be on record as saying: >>> "SQLite can be thought of as a derivative of PostgreSQL. SQLite was >>> originally written from PostgreSQL 6.5 documentation, and the SQLite >>> developers still use PostgreSQL as a reference platform to verify that >>> SQLite is working correctly." >>> >>> Not a major problem for me, just an interesting footnote. >>> >> The way I recall hearing it expressed in talks I've watched (recorded > after >> the fact, never live) is that when situations arise that need resolution, > the >> question is often asked "what does PostgreSQL do?" Sadly, the backward >> compatibility requirements prohibit SQLite from being a 100% feature / >> implementation match. > Absolutely so. Nevertheless, it would be pretty minor to add LIMIT ALL as a > synonym for LIMIT -1. If you wanted to enhance compatibility.
It would also be very minor to add "UNION -1" as a synonym for "UNION ALL", but "being minor" is not an argument for doing so. While the mentorship of Postgres is undoubted, there is/was never a drive, nor a need for full (or even partial) compatibility with "Postgres" per se, mostly care is applied to conform or be compatible with the SQL standard as much as possible (much like PostGres' philosophy) - At least this is how I read the Dev's statements thus far. If you can show that the SQL standard likes the "LIMIT ALL" phrasing, or argue that it has in it's own right an advantage over "LIMIT -1", then you would have a much better case than just saying "But Postgres does it", and then it would make sense even if it isn't very minor. All that said, personally I do like the "LIMIT ALL" for clarity and wouldn't mind seeing it implemented. Cheers, Ryan
