This reminds me of nothing more than the "joys" of working with Splunk. They have a query language with a big fat manual and a ton of online documentation. What they
lack is a query language that is internally consistent, has an actual proper definition with a grammar, etc. In some functions, arguments that involve fields whose name includes a '.' (one character dot/period/full-stop) must be quoted with single quotes. In other functions, those same arguments must be quoted with
double quotes. In yet other functions, those exact same arguments must be left without any quotes.
Getting any of those cases wrong doesn't yield an error message either! Instead, it gives you wrong output, where "wrong" ranges between incomplete, empty, and just-plain-crazy. When you go read the manual for the various functions, they don't bother mentioning any of this.
[Of course, this is only one minor example out of a giant ant-farm of similar bug-inducing inconsistencies!]
I have the "honor" to work with the largest single Splunk cluster known to exist (there may be other larger ones that are in gov't black programs or are unmentionable for some unknown reason). So we twisted the tech rep's arms
really hard, and finally found out why things are so inconsistent. It seems that different functions and areas of functionality were implemented at different times, by different teams, with hand-written parsers and absolutely no standards whatsoever. There's no grammar available for their query language
because they can't actually tell what it is without reading and reverse-engineering all the parsing code! 
And, of course, now that they have a bazillion customers they can't
change any of that either, because they'd be certain to break some important customers' stuff.
