Random notes on query languages.

I started to include this in my post concerning my new language, but it relly isn't very helpful to the article. I want to keep the information around for myself later. Not all of these are generic information query languages, but all are examples of languages designed to ask for something.

QL :
Compiles to SQL. A combination of Datalog and SQL. Looks like it has a form of list comprehension. That could be useful.

Common Query Language :
This seems to be an attempt to standardize what thed user has to type into the innumerably implemented searches online and off. Simplicity is always nice to consider.

D :
Apparently this was a truly relational language. Google tells me concerning the site pointed at for it from wikipedia, "This site may harm your computer". I think this one best left alone.

Data Definition Language :
This appears to be a fuzzy logic belief system buried in a syntax molded to look like SQL.

Datalog :
Logic programming. ancestor(X,Y) := Parent(X,Y). The ancestor of X is Y if the Parent of X is Y. Could be useful. Looks rather constraining in what it could do easily.

This just led me to read about entity-relationship models, finally giving name to a project I created out of frustration with data manipulation about a year ago. This is a great system for data, this particular language is a natural language version interface. Better for people that machines to query with.

I found references to this but could not locate an example of its syntax.

This isn't a query language I hear you saying. Well, you're right. But I'm looking at data query, and this does query data, albeit not relational. The manner LDAP uses to filter datasets is nice. Very combinable, readable and easily created programatically. On being nice, forcing everything into a strict hierarchy is not.

Not so much a language as a search API for finding structures and substructures in molecular simulations.

MDX : MultiDimensional eXpressions
This appears to have syntax borrowed from SQL with the addition of prepartitioned data and some niceties in reaching it. There are some interesting ideas at work here, I will have to look into them in the future.

The binding of SQL tables and their associated columns to devices and their exposed attributes. The interesting difference is that instead of just returning scalars it can also return objects that can be traversed using dot-member syntax. The aggregation method differs from standard SQL in dividing items into groups that can then be referenced by values and subqueries using the keyword partition.

OPath :
A combination of API and string segment arguments that allow for locating items in the WinFS store. Apparently there will be a natural language to OPath translator. It looks like you acquire a query-object into a potion of the WinFS store programatically and then pass it a string specifying the boolean operations to indicate which items from the current partition of the store are desired.

SQL with a more complex from statement involving manipulating ranges and dropping the select keyword.

This is a language designed for querying information found in RDF files. I applaud them for not merely repurposing the syntax of SQL to the task. It seems that many think this is what creating a query language means. The interesting part of the query consists of the where clause. Apparently the items returned are generated by specifying a series of interconnected assertions that whenever met return a clump of information consisting of all the specified variables that where intermixed with the assertions being bound to the values that allowed the assertions to match. It looks neat, but it bothered me that just glancing at it the query did not produce a mental model of how it operated. Maybe it was just me.

The current cream of the crop in relational data manipulation, SQL has been allowing us to keep an external dataset and query and alter it in standard ways for more than thirty years. Of course the primary part that seems to be standard is mostly in the fashion that queries are marked up and some general agreement on what terms to use to indicate the various ways of stitching together tables of data. In general anything so clever as to be particularly interesting or useful is custom to only a given database and the few that might choose to copy them. For a standard language, the implementation specific libraries do much to unstandardize it. The primary limitation of the language in my experience is that anything you want to do must be expressed through the metaphor of stitching together tables of information and of course returning a single table of information resulting from this. Many clever hacks are used to bypass these limitations in practice of coarse, but trying to do anything other than simple column binds can become very confusing to look at very quickly.

Microsofts query only SQL for WMI. A hidden SQL server on every server and workstation of a domain that only understands select queries. Useful.

No comments:

About Me

(2) the subculture of the compulsive programmer, whose ethics prescribe that one silly idea and a month of frantic coding should suffice to make him a life-long millionaire. --ewd1036