Disable Spring by default when running executable - #301

[soberstadt]

I don't know very much about rake tasks, so I don't know how to fix if it is running with rake, but this fixes bundle exec parallel_rspec.

[grosser]

this only fixes rspec :)

I'd say put it intoParallelTests::CLI#run ... ideally with a test that assets that it is getting set

[soberstadt]

Ah, I didn't realize this was an issue on all of the suites. I'll look into that.

[soberstadt]

I updated the code to set the var in the CLI class, and tested it in practice and it works for both the executable and the rake task. But now I am struggling to understand your tests and where I should check if it was set. The test I wrote won't pass.

[grosser]

I can fix up the tests, thx for verifying :)

[soberstadt]

Thanks for encouragement and contributions in OSS!

[grosser]

2.2.0 is out! :D

[betesh]

This breaks all my tests. I was using parallel_tests with spring and everything was working fine with a little help from #309 (comment)

I don't see any documentation warning uses about this breaking change, and I don't see from the code how it's possible to override it.

[grosser]

why is it breaking your tests ?
can you post some backtraces or instructions to reproduce ?

The 309 patch is hard to generalize since it would need to reconnect all
your data sources (redis/memcached etc) ... it's a nice hack for super fast
tests though ... so I like the idea of supporting it ... for example by
doing DISABLE_SPRING||=1 and you could set it to '' or '0' to prevent that
override ...

On Wed, Dec 2, 2015 at 11:03 AM, betesh notifications@github.com wrote:

This breaks all my tests. I was using parallel_tests with spring and
everything was working fine with a little help from #309 (comment)
#309 (comment)

I would think if a default is going to change, there's at least be a
documented way to override it.

—
Reply to this email directly or view it on GitHub
#468 (comment)
.

[grosser]

PR #469

On Thu, Dec 3, 2015 at 8:28 AM, Michael Grosser michael@grosser.it wrote:

why is it breaking your tests ?
can you post some backtraces or instructions to reproduce ?

The 309 patch is hard to generalize since it would need to reconnect all
your data sources (redis/memcached etc) ... it's a nice hack for super fast
tests though ... so I like the idea of supporting it ... for example by
doing DISABLE_SPRING||=1 and you could set it to '' or '0' to prevent that
override ...

On Wed, Dec 2, 2015 at 11:03 AM, betesh notifications@github.com wrote:

This breaks all my tests. I was using parallel_tests with spring and
everything was working fine with a little help from #309 (comment)
#309 (comment)

I would think if a default is going to change, there's at least be a
documented way to override it.

—
Reply to this email directly or view it on GitHub
#468 (comment)
.

[betesh]

That should work--I'll give it a shot tomorrow.

The backtraces wouldn't really tell you anything--it's just clear from them that different processes are using the same database because I get a lot of Mysql2 errors complaining about UNIQUE INDEX constraints. When I run the same suite with parallel_tests but without spring, everything passes.

This reminds me of another suggestions--it would be great if when running tests (rspec in my case), the output would mention the TEST_ENV_NUMBER somewhere, and maybe also the DB name. Maybe a custom RSpec formatter would work for this, but I suspect there's a more general solution that will work for any test framework.

It would give the user a quick verification that everything is working correctly--running in different processes, with different DB's.

[grosser]

why would they use the same database if spring was disabled ?

running with --verbose should show the TEST_ENV_NUMBER ... showing the
database is a bit trickier and needs some sort of AR integration

On Thu, Dec 3, 2015 at 4:09 PM, betesh notifications@github.com wrote:

That should work--I'll give it a shot tomorrow.

The backtraces wouldn't really tell you anything--it's just clear from
them that different processes are using the same database because I get a
lot of Mysql2 errors complaining about UNIQUE INDEX constraints. When I run
the same suite with parallel_tests but without spring, everything passes.

This reminds me of another suggestions--it would be great if when running
tests (rspec in my case), the output would mention the TEST_ENV_NUMBER
somewhere, and maybe also the DB name. Maybe a custom RSpec formatter would
work for this, but I suspect there's a more general solution that will work
for any test framework.

It would give the user a quick verification that everything is working
correctly--running in different processes, with different DB's.

—
Reply to this email directly or view it on GitHub
#468 (comment)
.

[betesh]

why would they use the same database if spring was disabled ?

This is a good question, and I don't know the answer. It's especially odd considering then when I just don't use spring at all, tests pass but when I run the command with spring and let parallel_tests disable it, tests fail.

Maybe by the time parallel tests is loaded, some of spring has already been loaded and executed and only part of its behavior ends up being disabled. That is wild speculation on my part but nothing else comes to mind.

Regardless, I just tested with #469 and it works.