Current state of things regarding ActiveRecord bind variable support

rubyonrails-core
#1

Hey all-

So last I posted, AR was looking into merging ARel support, and I was
playing with bind var support in my rails branch, mainly for Oracle.
Due to the uncertainty of the ARel merge, I stepped back a bit from my
git clone of miloops/rails since so much was changing.

Bottom line, though, I really need bind var support for AR at this
point, so I want to step back up and contribute however I can.

Also, I want to just throw this out: since I'm head of online dev at
PlayStation SD, I can help fund efforts for people that are going to
help out with this effort. So if any Rails core members are
interested in earning a few $K (Josh?? Jeremy??), I can provide
funds.

All that said, here's the point of my post/questions:

   - Is ARel still going to be merged into Rails in some form? How
close/far is that? What are the stopping blocks?
   - Looking thru the ARel patch/merge job vs, say, the Sequel class
hierarchy for expressions, Sequel is structured much more cleanly.
Take a look at sequel/dataset/sql.rb - it's really nicely divided.
Can we copy that?
   - Is there any reason the AR/ARel merge is not being majorly
refactored other than time/$$? When I see AREL = AR in the base
class, that tells me the whole thing is backwards. I'm not trying to
throw stones - my team is heavily dependent on AR, so getting the
best, most efficient product is VERY important.

Again, I can throw money at this problem for those that are
interested. I'm not asking for undying allegiance to PlayStation in
return - just that we get a more efficient AR that supports bind vars.

Anyways, if you could let me know what the status of the AR/ARel merge
is currently, that would help. Feel free to respond privately if
that's appropriate.

Thanks
Nate [nate playstation sony com]

#2

Anyways, if you could let me know what the status of the AR/ARel merge
is currently, that would help. Feel free to respond privately if
that's appropriate.

The intention is still to merge arel for 3.0, however I'm slightly out
of the loop on how it's progressing. I've pinged emilio about this,
so hopefully he'll pipe up soon.

#3

Hey all-

So last I posted, AR was looking into merging ARel support, and I was

playing with bind var support in my rails branch, mainly for Oracle.

Due to the uncertainty of the ARel merge, I stepped back a bit from my

git clone of miloops/rails since so much was changing.

Bottom line, though, I really need bind var support for AR at this

point, so I want to step back up and contribute however I can.

Also, I want to just throw this out: since I’m head of online dev at

PlayStation SD, I can help fund efforts for people that are going to

help out with this effort. So if any Rails core members are

interested in earning a few $K (Josh?? Jeremy??), I can provide

funds.

All that said, here’s the point of my post/questions:

  • Is ARel still going to be merged into Rails in some form? How

close/far is that? What are the stopping blocks?

Hi Nate,

Last night Arel integration into ActiveRecord was finally merged to rails/master. There will be bugs and other things to take care of before having a “stable” version (aka Rails 3) but we are getting there.

After the merge all test are passing for MySQL, SQLite3 and PostgreSQL, the 3 major supported databases.

  • Looking thru the ARel patch/merge job vs, say, the Sequel class

hierarchy for expressions, Sequel is structured much more cleanly.

Take a look at sequel/dataset/sql.rb - it’s really nicely divided.

Can we copy that?

Arel::Relation shouldn’t include the SQL logic, this is divided in file’s hierarchy “arel/engines/sql/relations” but not in classes. That would be not extending Relation’s with SQL specific logic, and be able to use it with any other engine (XML, json, etc).