Prevent calling same class methods inside of schema files

[stloyd]

Change Log

Added

Fixed

Changed

• Prevent calling same class methods inside of schema files

Removed

Deprecated

Security

---

Description

Skipping calling additional methods in the matter of accessing internal variables in class improves performance.

See: https://3v4l.org/sOIct

[norberttech]

Hmm not sure if I get the intention of this, is that affecting performance in any way?

[github-actions]

Flow PHP - Benchmarks

^{_{Results of the benchmarks from this PR are compared with the results from 1.x branch.}}

Extractors

+-----------------------+-------------------+------+-----+-----------------+--------------------+
| benchmark             | subject           | revs | its | mem_peak        | mode               |
+-----------------------+-------------------+------+-----+-----------------+--------------------+
| AvroExtractorBench    | bench_extract_10k | 5    | 3   | 44.121mb        | 517.847ms          |
| CSVExtractorBench     | bench_extract_10k | 5    | 3   | 14.006mb        | 435.235ms          |
| JsonExtractorBench    | bench_extract_10k | 5    | 3   | 18.667mb        | 823.888ms          |
| ParquetExtractorBench | bench_extract_10k | 5    | 3   | 237.791mb       | 1.153s             |
| TextExtractorBench    | bench_extract     | 5    | 3   | 3.935mb +51.22% | 3.189ms +74010.70% |
| XmlExtractorBench     | bench_extract     | 5    | 3   | 3.935mb +21.65% | 1.788ms +49304.36% |
+-----------------------+-------------------+------+-----+-----------------+--------------------+

Transformers

+-------------------------------------+----------------------------+------+-----+----------------+------------------+
| benchmark                           | subject                    | revs | its | mem_peak       | mode             |
+-------------------------------------+----------------------------+------+-----+----------------+------------------+
| EntryExpressionEvalTransformerBench | bench_transform_json_row   | 1000 | 5   | 3.935mb +0.75% | 26.538μs +43.90% |
| EntryExpressionEvalTransformerBench | bench_transform_string_row | 1000 | 5   | 3.935mb +0.75% | 19.546μs +29.90% |
| EntryExpressionEvalTransformerBench | bench_transform_xml_row    | 1000 | 5   | 3.935mb +0.75% | 58.169μs +57.90% |
| RenameEntryTransformerBench         | bench_transform            | 1000 | 5   | 3.935mb +0.76% | 6.324μs -74.35%  |
+-------------------------------------+----------------------------+------+-----+----------------+------------------+

Entry Factory

+-------------------------+----------------+------+-----+----------+-----------+
| benchmark               | subject        | revs | its | mem_peak | mode      |
+-------------------------+----------------+------+-----+----------+-----------+
| NativeEntryFactoryBench | bench_10k_rows | 5    | 5   | 83.901mb | 185.339ms |
| NativeEntryFactoryBench | bench_1k_rows  | 5    | 5   | 45.446mb | 17.297ms  |
| NativeEntryFactoryBench | bench_5k_rows  | 5    | 5   | 62.730mb | 91.050ms  |
+-------------------------+----------------+------+-----+----------+-----------+

[stloyd]

@norberttech It was late when I pushed this 😂 Added comment into description:

Skipping calling additional methods in the matter of accessing internal variables in class improves performance.
See: 3v4l.org/sOIct

[norberttech]

I would rather postpone this adjustment for now, there are two reasons:

• it does not seem to be impactful enough to bring a visible value to the performance at this state
• For some of those methods I'm planning to make mnemonics that will execute only once (like flatPath), and then return the result like, for example isList or isListElement, that's why I'm using methods directly rather than properties