when using rbspy, anyone seen a trace like this:
Time since start: 86s. Press Ctrl+C to stop.
Summary of profiling data so far:
% self % total name
99.74 100.00 <c function> - unknown
0.02 0.02 extract_record - /Users/kbrock/.gem/ruby/2.5.5/gems/activerecord-5.1.7/lib/active_record/associations/join_dependency/join_part.rb
0.01 100.00 instrument - /Users/kbrock/.gem/ruby/2.5.5/gems/activesupport-5.1.7/lib/active_support/notifications/instrumenter.rb
0.01 100.00 block in instantiate - /Users/kbrock/src/gems/activerecord-virtual_attributes/lib/active_record/virtual_attributes/virtual_fields.rb
0.01 100.00 block (2 levels) in instantiate - /Users/kbrock/src/gems/activerecord-virtual_attributes/lib/active_record/virtual_attributes/virtual_fields.rb
0.01 99.83 calculate - /Users/kbrock/src/gems/activerecord-virtual_attributes/lib/active_record/virtual_attributes/virtual_fields.rb
0.01 99.81 perform_calculation - /Users/kbrock/.gem/ruby/2.5.5/gems/activerecord-5.1.7/lib/active_record/relation/calculations.rb
0.01 0.04 visit - /Users/kbrock/.gem/ruby/2.5.5/gems/arel-8.0.0/lib/arel/visitors/reduce.rb
but only 1k rows and it is quick from psql
I'll reinstall the gems
Hi Everyone, While starting the evm by 'bundle exec rake evm:start' I am getting
'Base::EventCatcher' as Aborted and the workers gets increased tremendously causing the evm to run really slow. Kindly help. However the evm gets started but with around 20000 workers in which half of them are
Base::EventCatcher : Aborted
'Base::EventCatcher' as Aborted and the workers gets increased tremendously causing the evm to run really slow. Kindly help. However the evm gets started but with around 20000 workers in which half of them are
Base::EventCatcher : Aborted
Think someone here mentioned https://github.com/foraker/rspec_profiling
I haven't had a chance to check it out, but does look nice
article: https://www.foraker.com/blog/profiling-your-rspec-suite
I haven't had a chance to check it out, but does look nice
article: https://www.foraker.com/blog/profiling-your-rspec-suite
@agrare I was chatting with @Fryguy about some things I'm finding looking at memory profiling of manageiq...You might want to see this
-"joe" notation in ruby 2.5+ that and the String#clear discussed here: https://rubytalk.org/t/psa-string-memory-use-reduction-techniques/74477
It looks like the
-"joe" notation for deduplicating strings was added in many places for rails 6.0
and then String#clear could be used where we do big XMLs, remove the reference and then GC
I don't think that's for VImStrings though
I did a string pool for things like that ages ago, but then we changed it to symbols
since symbols did the same task memory-wise
even so there are still some string savings there
like in the vimmappingregistry I think
regardless, wanted this on your radar
...
In other words, there’s no benefit in writing any of the following:
foo = { "key".freeze => nil } # unnecessary freeze
foo["a".freeze] = true # unnecessary freeze
foo["b".freeze] # unnecessary freeze
They are equivalent to the following, in all versions of Ruby:
foo = { "key" => nil }
foo["a"] = true
foo["b"]
namely, each call to
foo["a"] doesn't allocate a string for a hash class but I wonder about VimHash
oh, I forgot about that @Fryguy, I don't remember doing that
:laughing:
we might get some savings by putting the frozen string literal pragma on the top of the routes.rb files but I need to measure it