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

5th
Feb 2015
Ken Collins
@metaskills
Feb 05 2015 12:41
Morning @sgrif
Nice twitter back and forth yesterday on core work.
Ken Collins
@metaskills
Feb 05 2015 12:52
I got my own Azure instance setup last night and will be fininising some TinyTDS enhancements and making sure the adapter works.
Sean Griffin
@sgrif
Feb 05 2015 14:12
Ha
That conversation made me angry
Ken Collins
@metaskills
Feb 05 2015 14:12
Yea, I can see both sides.
Sean Griffin
@sgrif
Feb 05 2015 14:13
The tweet about full-time being harmful from David made all the pushing I've done feel cheapened
Ken Collins
@metaskills
Feb 05 2015 14:13
Well, the DHH side, I think… is that we want to avoid super big $ thrown at it. Like what appears to be nodes issue.
And yes, that was bad.
Backreading again...
Sean Griffin
@sgrif
Feb 05 2015 14:14
Sure, I get that
It just felt hypocritical given how much Basecamp has directly profited from Rails
I'd argue he owes the majority of his financial success to the success of Rails, so arguing that we should be able to pay our bills as a result of our OSS work is aggravating
Ken Collins
@metaskills
Feb 05 2015 14:16
I remember when he went on about gittip and pissed off substack.
Who mades a moderate living from it.
So yea, agreed, there are many many more people doing moderately well.
And sometimes he seems to poo poo on em.
Sean Griffin
@sgrif
Feb 05 2015 14:31
Btw managed to get this done. Hoping to do the same on MRI. jruby/jruby#2562
Ken Collins
@metaskills
Feb 05 2015 14:55
WHOA.
Ken Collins
@metaskills
Feb 05 2015 19:29
Picking up on The Bike Shed again today.
Today, 3: Flipping the Script
FYI, my mocha-phantoms NPM will need to drop CoffeeScript due to PhantomJS 2.0
Ken Collins
@metaskills
Feb 05 2015 19:48
@sgrif I really like this. Never listened to pod casts.
Ken Collins
@metaskills
Feb 05 2015 20:06
Re the arity of a block, AR solves that in many places… it just uses *block_args and splats them back.
rahsya
@rahsya
Feb 05 2015 20:26
@metaskills Hi, I'm a developer supporting a Rails site that greatly benefits from the tiny_tds and activerecord-sqlserver-adapter gems; so thanks for your work first of all!
Ken Collins
@metaskills
Feb 05 2015 20:26
Word!
Glad you are finding it useful.
rahsya
@rahsya
Feb 05 2015 20:28
I've been following a couple issue/bug fix threads on GitHub that have been super useful in debugging some issues we've encountered in production
you happened to be in all of them so it's pretty cool I can thank you in real time :)
anyways happened to have a question about your recent 4.2 release of the activerecord-sqlserver-adapter
Ken Collins
@metaskills
Feb 05 2015 20:30
Nice, yea, I have been closing issues with a passion.
Sure.
Sean Griffin
@sgrif
Feb 05 2015 20:30
Sure, but splatting the args is just as expensive
It really only mattered because read_attribute is pretty much the definition of a performance hotspot
Ken Collins
@metaskills
Feb 05 2015 20:31
Whoa, that’s great to know.
Sean Griffin
@sgrif
Feb 05 2015 20:32
I shouldn't say just as expensive.
It could be cheaper
But it has a cost
And if I am at the point where the cost of &block is a bottleneck, I can't afford to pay the *args cost either
The better solution is just to fix it in the compiler. XD
Ken Collins
@metaskills
Feb 05 2015 20:32
Noted, I’ve been a "yield to the block when given” and “block_args” proponent for some time.
rahsya
@rahsya
Feb 05 2015 20:33
The TLDR version is the problem we are facing right now in production is an issue that existed in tiny_td 0.6.1 which is occasionally the app doesn't detect the lose of a connection to an Azure SQL DB and gets stuck in a connection retry loop
( for context: rails-sqlserver/tiny_tds#138)
upgrading to 0.6.2 fixed that problem but there was still an issue where after a period of time the connection to Azure which would go stale and the next query would timeout (for context: rails-sqlserver/tiny_tds#164)
Ken Collins
@metaskills
Feb 05 2015 20:36
@rahsya rails-sqlserver/tiny_tds#162
I am working on that issue over the next few days.
Wow, really like how Gitter uses red/green to indicate closed/open
rahsya
@rahsya
Feb 05 2015 20:37
yeah that is nice - actually I'm new Gitter altogether and man this is actually pretty great
Ken Collins
@metaskills
Feb 05 2015 20:38
So your issue goes away when that one does right?
Also, the 4.2 adapter (if you can use it) does a much better job of leveraging #active? than others did.
rahsya
@rahsya
Feb 05 2015 20:39
ok great yeah that's exactly what I'm planning on doing right now - upgrading our gem to 4.2
Ken Collins
@metaskills
Feb 05 2015 20:39
Remember, 4.2 requires SQL Server 2012 or 2014.
rahsya
@rahsya
Feb 05 2015 20:40
yep we're running Azure SQL Server 2012
Ken Collins
@metaskills
Feb 05 2015 20:40
You just made Fetch happen!
Sean Griffin
@sgrif
Feb 05 2015 20:40
<_<
rahsya
@rahsya
Feb 05 2015 20:42
Yeah so looks like 162 is similar to what we're experiencing
Although it's not a very consistent 5 mins, we end up getting timeouts inconsistently at variable times
Ken Collins
@metaskills
Feb 05 2015 20:44
Well, those are hard to diagnose as TinyTDS problems
It could easily be just query timeouts and performance of the platform and what tier you are running on.
rahsya
@rahsya
Feb 05 2015 20:47
Yeah definitely - there are application-specific things I know may be confounding the timeouts we're experiencing but actually simply upgrading from 0.6.1 -> 0.6.2 of tiny_tds solved most of the timeout issues we experienced
Anyways I'll start following the thread for rails-sqlserver/tiny_tds#162
And try to test any fixes that may come out of that
My original question actually was: in the 4.2 release of activerecord-sqlserver-adapter some of the commits suggest that the "connection retry" handling code was removed
rahsya
@rahsya
Feb 05 2015 20:54
Was that removed due to issues that people have had? Or a decision that it wasn't the responsibility of the gem?
oh looks like you provided context in this commit here :) :rails-sqlserver/activerecord-sqlserver-adapter@a347f55
Ken Collins
@metaskills
Feb 05 2015 21:00
Yea, that does sum it up :)
Ken Collins
@metaskills
Feb 05 2015 22:29
Signing off.