[2019-06-20]
javiereguiluz [10 h 07]
There is not an automated process to track the performance improvement/regression of each pull request
the main reason is that it's difficult and very time consuming to create the needed performance benchmarks. Each component (and each feature inside each component) would need a dedicated benchmark
For example, you make a minor change in OptionsResolver component ... a generic benchmark using forms, etc. says that your change worsened performance a lot, and your PR is rejected.  However, a benchmark related to Forms is testing forms, optionsresolver and lots of other things. The 5 ms added in the benchmark by your pull request, are in fact randomly added by some unrelated forms feature
So, you need a dedicated synthetic benchmark for each feature of each component

javiereguiluz [10 h 17]
(in any case, we could create some benchmarks for features where performance is ultra critical, such as getting services/params from container, things related to request, etc.)

Noni [10 h 25]
@javiereguiluz thank you for your time and answer
in fact i was talking about your second post yes, mainly core components
like http request/response (parameter bags, factories)
dependencie injection, console, event dispatcher, and some others key component

also what do you think of proposing such improvement in the software release process ?
obviously not in an automated way on the PR, but in a post merge process ? in preparation for LTS for example

the community could of course do this as well, but maybe need some documentation on using blackfire or other tooling code :visage_légèrement_souriant:

anyway, thank you !

javiereguiluz [10 h 28]
Personally I think it's a reasonable proposal ... but I don't know what would the Symfony Core think about this. So, would you be willing to create an issue at https://github.com/symfony/symfony/issues explaining briefly this proposal and see what happens? Thanks!

Noni [10 h 31]
lets do this :visage_légèrement_souriant: