These are chat archives for rails-sqlserver/activerecord-sqlserver-adapter

26th
Jan 2015
One down
Been refactoring towards that for weeks, feels good to finally be able to pull it off
Ken Collins
@metaskills
Jan 26 2015 00:53
WOW, congrats!
Just taking a quick look...
Sean Griffin
@sgrif
Jan 26 2015 00:53
TL;DR, no more where_values
Ken Collins
@metaskills
Jan 26 2015 00:56
This looks sharp!
Sean Griffin
@sgrif
Jan 26 2015 00:56
Still need to move having and joins over to this structure, (which requires unravelling some more rats' nests)
But it's progress
Sean Griffin
@sgrif
Jan 26 2015 01:14
Random question: If I replaced all of these test descriptions (https://github.com/rails/rails/blob/master/activerecord/test/cases/relation/where_clause_test.rb#L5-L29) with "WhereClause is a monoid", do you think most developers would understand, or am I over-estimating the number of people who have seen/been influenced by Haskell/category theory
Ken Collins
@metaskills
Jan 26 2015 01:30
Reading...
Something to share too… rails-sqlserver/activerecord-sqlserver-adapter#376
I never took computer science. Nor did much digging in the meaing of a monoid.
Sean Griffin
@sgrif
Jan 26 2015 01:33
It's more of a math term than a CS term
I also didn't take computer science. :)
Ken Collins
@metaskills
Jan 26 2015 01:33
Heck yea!
Sean Griffin
@sgrif
Jan 26 2015 01:34
Best method name
make_Fetch_Possible_And_Deterministic
Ken Collins
@metaskills
Jan 26 2015 01:34
Intention revealing design :)
With flair!
One more issue to doc, then to make a release article. Finished with a pre gem push.
Sean Griffin
@sgrif
Jan 26 2015 01:35
One thing that stands out to me: isn't determinism of the query separate from the primary key?
e.g. Isn't it deterministic as long as it's ordered by any column with a unique index?
If I'm reading this correct, the direction you're heading is that this belongs on the connection adapter, rather than the relation, right? If so I agree
Also I have no clue why we're removing the order. My guess is some nonsense with MySQL? If so we should probably move the removal of the order clause over to the connection adapter
Ken Collins
@metaskills
Jan 26 2015 01:38
It could totally be that this is just a bad test. I hate that other DBs give you an illusion of PK ordering.
Will work it out later.
Sean Griffin
@sgrif
Jan 26 2015 01:41
You mind reviewing this talk proposal for me?
Ken Collins
@metaskills
Jan 26 2015 01:51
I do not!
Type Hard - Keep Being Type Casted
Sean Griffin
@sgrif
Jan 26 2015 01:55
Rofl
Ken Collins
@metaskills
Jan 26 2015 01:56
or they're too stupid to understand it
I would break that up a bit so it reads better. When scanning, it could sound wrong.
I really like the overview. I’ve learned a lot watching your approach.
Small responsible objects FTW.
Sean Griffin
@sgrif
Jan 26 2015 01:57
I was thinking of rewording it as "they think they're not smart enough to understand it" or just as "it goes over their heads"
Ken Collins
@metaskills
Jan 26 2015 01:57
Yes.
Sean Griffin
@sgrif
Jan 26 2015 01:57
Which one do you think is better?
Feel free to be brutal, btw. I know I'm shit at getting my intentions across
Ken Collins
@metaskills
Jan 26 2015 01:58
First.
Ditto.
Sean Griffin
@sgrif
Jan 26 2015 01:59
I gotta run, but I'll read any other feedback in a few hours
Ken Collins
@metaskills
Jan 26 2015 02:01
TTYL!
Sean Griffin
@sgrif
Jan 26 2015 16:44
I moved the proposal to a PR if you have any more feedback: sgrif/talk-proposals#2