News for hackers. Stuff that matters.
Re:PostgreSQL 10 Beta 1 Released by ubertaco on Thursday May 18, 2017 @11:19AM (#14367946)
Or they want to allow for case-insensitivity of some data, like for example email addresses on login forms.

As much as postgres is overall better than MySQL in so many ways, it's still ridiculously difficult to set things up such that

    SELECT id FROM users WHERE email='[email protected]' 
returns the same result as

    SELECT id FROM users WHERE email='[email protected]'
Re:PostgreSQL 10 Beta 1 Released by rarrrrrr on Thursday May 18, 2017 @11:32AM (#14368074)
Here's an example of doing that in PostgreSQL:

  create table user_email (
      email text not null 
  );
  -- create a index on the lowercase form 
  -- of the email 
  create unique index user_email_case_idx 
      on user_email (lower(email));
  -- select using the index, with the lowercase form.
  select 1 
    from user_email 
   where lower(email)=lower('[email protected]');
Re:PostgreSQL 10 Beta 1 Released by olavgg on Friday May 19, 2017 @04:31AM (#14373809)
actually it is easier

SELECT 1 FROM user_email WHERE email ILIKE '[email protected]';

Re:PostgreSQL 10 Beta 1 Released by always_good on Friday May 19, 2017 @08:20AM (#14374867)

    WHERE lower(email) = lower('[email protected]')
Is simple and hits an index on lower(email).

I'm not sure ILIKE can hit an index in your example.

Re:PostgreSQL 10 Beta 1 Released by olavgg on Friday May 19, 2017 @03:37PM (#14378765)
It can if you use the pg_trgm extension, a good summary can be read here: https://niallburkley.com/blog/index-columns-for-like-in-post...
Re:PostgreSQL 10 Beta 1 Released by grahamedgecombe on Thursday May 18, 2017 @11:35AM (#14368116)
The citext type automatically does case-insensitive comparisons: https://www.postgresql.org/docs/current/static/citext.html
Re:PostgreSQL 10 Beta 1 Released by synotna on Thursday May 18, 2017 @11:27AM (#14368025)
Re:PostgreSQL 10 Beta 1 Released by ncd on Thursday May 18, 2017 @11:33AM (#14368087)
Postgres also has the citext column type to make this a snap.

https://www.postgresql.org/docs/9.6/static/citext.html

Re:PostgreSQL 10 Beta 1 Released by metalliqaz on Thursday May 18, 2017 @11:58AM (#14368340)
I'm no Postgres master by any means, but I searched it: https://duckduckgo.com/?q=postgres+case+insensitive+query

solution immediately came up at SO:

    SELECT id FROM users WHERE LOWER(email)=LOWER('[email protected]')
Re:PostgreSQL 10 Beta 1 Released by inopinatus on Thursday May 18, 2017 @04:02PM (#14370625)
That's a bad practice. Did you know: email addresses are case-sensitive on the left-hand-side.

It's discouraged by RFC5321 whilst also being defined by it.

Re:PostgreSQL 10 Beta 1 Released by fphilipe on Thursday May 18, 2017 @11:29AM (#14368041)
Does MySQL do that on varchar by default?

Can't you just do this in PostgreSQL?

    SELECT id FROM users WHERE email = lower('[email protected]')
Re:PostgreSQL 10 Beta 1 Released by thejosh on Thursday May 18, 2017 @11:44AM (#14368196)
Because you don't always want to match 'Foo' to 'foo'?
Re:PostgreSQL 10 Beta 1 Released by jakobegger on Thursday May 18, 2017 @01:40PM (#14369310)
That's what per-column collations are for. Ideally you should be able to choose from case sensitive and case insensitive collations. Unfortunately PostgreSQL doesn't support case insensitive collations (for some reason the string comparison routines use memcmp as a tie-breaker when the collation says strings are equal).
Re:PostgreSQL 10 Beta 1 Released by irrational on Thursday May 18, 2017 @02:26PM (#14369741)
If you want to do case-insensitive for all languages you can do this:

1. first install the following (be sure to replace [your schema]:

CREATE EXTENSION pg_trgm with schema extension;

CREATE EXTENSION unaccent with schema extension;

CREATE OR REPLACE FUNCTION insensitive_query(text) RETURNS text AS $func$ SELECT lower([your schema].unaccent('[your schema].unaccent', $1)) $func$ LANGUAGE sql IMMUTABLE;

2. then in your query you can use:

where insensitive_query(my_table.name) LIKE insensitive_query('Bob')

Re:PostgreSQL 10 Beta 1 Released by openasocket on Thursday May 18, 2017 @03:09PM (#14370145)
That will not work for all languages. Look at https://www.w3.org/International/wiki/Case_folding for an explanation of why this problem is nontrivial. The lower function is sufficient: it handles case-folding properly, using the configured locale. Explicitly stripping accents can actually be the wrong choice depending on the locale
Re:PostgreSQL 10 Beta 1 Released by mgkimsal on Thursday May 18, 2017 @12:10PM (#14368429)
not sure why you got downvoted so much.

Everyone's answer is "just lowercase everything".

I'll respond just a bit:

1. You don't always have control over all the queries that have been written against your database.

2. You would probably lose the ability to use ORMs without a moderate amount of customization.

3. If you're migrating from a different database, you may have checksums on your data that would all need to be recalculated if you change case on everything stored.

4. Doing runtime lowercase() on everything adds a bit of overhead, doesn't it?

citext on postgresql seems a decent option - the citext docs even mention drawbacks of some of the other recommended options.

Re:PostgreSQL 10 Beta 1 Released by openasocket on Thursday May 18, 2017 @03:00PM (#14370044)
Is there a single ORM out there that doesn't support the lower() function? I googled "case insensitive search" + a couple ORMs and each of them could implement it as a one-liner.

And doing the runtime lower() on everything will generally not be slower than citext. If you look at the source for the citext comparison (https://github.com/postgres/postgres/blob/aa9eac45ea868e6dda...) you'll see it is internally converting the values to lowercase and comparing them. All it saves you is the overhead of a sql function invocation, and you'd have to do a lot of comparisons to make that difference measurable. But if you're doing a lot of those comparisons, unless you're just running the calculation on the same couple values over and over, the memory and disk latency will dominate performance, not the minimal overhead of the sql function invocation.

I agree you should probably use citext if you need case-insensitive unique or primary key values, but be aware of the drawbacks. https://www.postgresql.org/docs/current/static/citext.html

Re:PostgreSQL 10 Beta 1 Released by solidsnack9000 on Friday May 19, 2017 @12:15AM (#14372981)
> 4. Doing runtime lowercase() on everything adds a bit of overhead, doesn't it?

Maybe MySQL has special sauce for doing this comparison without lowercasing the query string? But there must be some overhead relative to exact search?

Re:PostgreSQL 10 Beta 1 Released by sametmax on Thursday May 18, 2017 @11:30AM (#14368052)
Just lowercase everything. Not that hard.
Re:PostgreSQL 10 Beta 1 Released by irrational on Thursday May 18, 2017 @02:29PM (#14369770)
That only works if you are only dealing with English. I've posted a comment with a solution that works across all languages.