Hi all
I’m trying to optimise my spec run time. I have 123 examples so far,
which run in ~4.2 seconds on average. But 116 of those will run in ~0.18
seconds. So, obviously, I only want to run the slow ones when I change
that code.
I’ve added adapter: :slow to the offending example group, which covers
the whole of exactly one file.
I’ve added this to the top of ./Guardfile:
ENV["GUARD"] = "true"
and this to my ./spec/spec_helper.rb:
RSpec.configure do |config|
config.filter_run(focus: true)
config.filter_run_excluding(adapter: :slow) if GUARD
config.run_all_when_everything_filtered = true
end
This is my terminal output:
=====
Running:
spec/bitcoin/data_access/satoshi/bdb_satoshi_wallet_repository_spec.rb
No examples matched {:focus=>true}. Running all.
No examples were matched. Perhaps {:adapter=>:slow} is excluding
everything?
0 examples: 100% |==========================================| Time:
00:00:00
Finished in 0.00028 seconds
0 examples, 0 failures
This is not what I expected config.run_all_when_everything_filtered = true but from the Relish docs it looks like I misunderstood.
Is this a bug, or a missing feature that could go into a future RSpec,
or something that would not make sense anyway (i.e.
run_all_when_everything_filtered really should only respect the tags
from filter_run)?
And more pragmatically, is there any way I can achieve what I want now,
i.e. have that one file’s examples run but only when I save it?
Cheers
Ash
–
http://www.patchspace.co.uk/
http://www.linkedin.com/in/ashmoran
I’m trying to optimise my spec run time. I have 123 examples so far,
which run in ~4.2 seconds on average. But 116 of those will run in
~0.18 seconds. So, obviously, I only want to run the slow ones when I
change that code.
I have a similar situation with the slow specs being integration specs
in a special directory. I tagged all those example groups and setup two
guards. The first is for unit tests and ignores all examples tagged
‘integration’ and doesn’t watch spec/integration. The second guard just
watches spec/integration and runs only examples tagged ‘integration’.
Haven’t used it much yet, though.
cheers,
Nikolay
On 30 Aug 2011, at 07:01, Nikolay Sturm wrote:
I have a similar situation with the slow specs being integration specs
in a special directory. I tagged all those example groups and setup two
guards. The first is for unit tests and ignores all examples tagged
‘integration’ and doesn’t watch spec/integration. The second guard just
watches spec/integration and runs only examples tagged ‘integration’.
Haven’t used it much yet, though.
I never thought of that! Yes, that could also work, probably better in
fact. It just involves running multiple Guard processes, although
there’s Terminitor[1] for that!
At this point I didn’t want to separate my specs into unit/ and
integration/ folders. One surprisingly major reason is that I live off
the RSpec TextMate bundle’s Alternate File feature to swap between lib/
and spec/ files. It also makes writing the Guard matchers a lot easier
(i.e. you don’t have to change anything). I may change that if I find I
often want to locate files by their nature.
Cheers
Ash
[1] terminitor | RubyGems.org | your community gem host
–
http://www.patchspace.co.uk/
http://www.linkedin.com/in/ashmoran
I never thought of that! Yes, that could also work, probably better in
fact. It just involves running multiple Guard processes, although
there’s Terminitor[1] for that!
A single guard process is enough, it will start all guards defined in
your Guardfile.
cheers,
Nikolay
–
“It’s all part of my Can’t-Do approach to life.” Wally
On 30 Aug 2011, at 14:45, Nikolay Sturm wrote:
A single guard process is enough, it will start all guards defined in
your Guardfile.
I did not know that! I’m still new to Guard, a recent convert from
Autotest. Thanks for the tip.
Cheers
Ash
–
http://www.patchspace.co.uk/
http://www.linkedin.com/in/ashmoran
On 30 Aug 2011, at 14:45, Nikolay Sturm wrote:
I never thought of that! Yes, that could also work, probably better in
fact. It just involves running multiple Guard processes, although
there’s Terminitor[1] for that!
A single guard process is enough, it will start all guards defined in
your Guardfile.
This just came in really handy. I’m helping out now and again with a new
Rails project using Spork and Guard. We moved some code out into lib/
and set up a separate guard 'rspec' to only monitor lib/ and
spec/lib/. Took a bit of jiggery-pokery to make it not use Spork and
Rails but it works now, and is as fast as RSpec runs get.
I’m probably late to the party with Guard, as I used AutoTest for years.
But for the benefit of anyone new to either: if you’re not doing
continuous testing, you should be*, and if you are, you should never
be happy with how fast they run. They can always be faster. Always.
Guard makes it much easier than AutoTest to tune the test runs, and does
other cool stuff too.
*It was David C. who said the exact same thing to me on this
list, what- 5 or 6 years ago? One of the best pieces of advice I ever
got. Funny how for all the new fangled technology we’ve had in the mean
time (and for all I’ve learnt since), I’m still using the same basic
development process I used back then
Ash
–
http://www.patchspace.co.uk/
http://www.linkedin.com/in/ashmoran
On 30 Aug 2011, at 14:45, Nikolay Sturm wrote:
I never thought of that! Yes, that could also work, probably better in
fact. It just involves running multiple Guard processes, although
there’s Terminitor[1] for that!
A single guard process is enough, it will start all guards defined in
your Guardfile.
Hi Nikolay
Just reviving this old thread as I ran into problems with multiple RSpec
guards[1] (patched now). Have you ever seen this behaviour? I forgot to
write an acceptance test so as yet nobody believes me it happens 
Ash
[1] Multiple Guard::RSpec instances don't run all files · Issue #74 · guard/guard-rspec · GitHub
–
http://www.patchspace.co.uk/
http://www.linkedin.com/in/ashmoran