[DavidBadura]

Renaming a field breaks the stored data, so far that needed an upcaster which moves the value from the old to the new key. NormalizedName accepts aliases now, similar to the event aliases in event-sourcing:

#[NormalizedName('email', aliases: ['mail'])]
public Email $email;

While hydrating, the first alias in the data is used if the field itself is missing. Extracting always writes the field name, so the stored data catches up over time. Field names and aliases have to be unique within a class and its parents, the metadata factory throws DuplicatedFieldNameInMetadata otherwise.

The ReflectionMapper only looks at the aliases if the field is missing. The generated mappers copy the alias into the field right before the assignment, classes without aliases get exactly the same code as before. The aliases are part of the fingerprint, and so is the property name now: renaming only the property behind an unchanged field name used to keep the outdated generated mapper.

Middlewares see the data as it is stored. The CryptographyMiddleware runs before the mapper, so it now also finds sensitive data and subject ids stored under an alias, otherwise a renamed sensitive field would not be decrypted.

#233 intersects the data with the fields kept by a property filter, which would drop the aliases. Whichever of both is merged second gets the fix for that.

[github-actions]

Hello 👋

Here are the most recent benchmark results, compared against 3.0.x. Time is the mode over the iterations in the its column, with the relative standard deviation in brackets.

GeneratedHydrator

subject	revs	its	memory	base	current	diff
benchHydrate1Object	1000	5	1.855mb -21.23%	0.789μs (±0.43%)	0.802μs (±2.80%)	+1.64%
benchExtract1Object	1000	5	1.855mb -21.23%	0.931μs (±0.67%)	0.939μs (±1.72%)	+0.81%
benchHydrate1000Objects	3	5	1.855mb -21.23%	765.890μs (±1.39%)	768.630μs (±0.90%)	+0.36%
benchExtract1000Objects	3	5	1.855mb 0.00%	488.763μs (±0.91%)	491.454μs (±0.77%)	+0.55%

StackHydrator

subject	revs	its	memory	base	current	diff
benchHydrate1Object	1000	5	1.855mb 0.00%	1.525μs (±0.73%)	1.544μs (±3.63%)	+1.26%
benchExtract1Object	1000	5	1.855mb 0.00%	1.925μs (±1.69%)	1.921μs (±5.83%)	-0.19%
benchHydrate1000Objects	3	5	1.855mb 0.00%	1.493ms (±0.95%)	1.505ms (±1.19%)	+0.81%
benchExtract1000Objects	3	5	1.855mb 0.00%	1.453ms (±0.62%)	1.443ms (±1.54%)	-0.67%

StackHydratorWithLazy

subject	revs	its	memory	base	current	diff
benchHydrate1Object	1000	5	1.855mb 0.00%	0.359μs (±2.60%)	0.343μs (±2.41%)	-4.49%
benchHydrate1ObjectTriggerInit	1000	5	1.855mb 0.00%	2.105μs (±4.07%)	2.118μs (±0.54%)	+0.61%
benchHydrate1000Objects	3	5	1.855mb 0.00%	332.256μs (±1.07%)	331.703μs (±0.52%)	-0.17%
benchHydrate1000ObjectsTriggerInit	3	5	1.855mb 0.00%	2.075ms (±0.97%)	2.105ms (±1.75%)	+1.43%

StackHydratorWithCryptography

subject	revs	its	memory	base	current	diff
benchHydrate1Object	1000	5	1.855mb 0.00%	2.431μs (±1.23%)	2.649μs (±0.52%)	+8.97%
benchExtract1Object	1000	5	1.855mb 0.00%	7.856μs (±2.78%)	7.854μs (±0.43%)	-0.02%
benchHydrate1000Objects	3	5	1.855mb 0.00%	2.389ms (±0.99%)	2.564ms (±0.78%)	+7.30%
benchExtract1000Objects	3	5	1.855mb 0.00%	7.227ms (±1.42%)	7.376ms (±0.68%)	+2.06%

GeneratedHydratorWithCryptography

subject	revs	its	memory	base	current	diff
benchHydrate1Object	1000	5	1.855mb 0.00%	1.695μs (±0.55%)	1.846μs (±0.68%)	+8.89%
benchExtract1Object	1000	5	1.855mb 0.00%	6.618μs (±4.11%)	6.543μs (±1.48%)	-1.13%
benchHydrate1000Objects	3	5	1.855mb 0.00%	1.634ms (±0.34%)	1.824ms (±0.81%)	+11.63%
benchExtract1000Objects	3	5	1.855mb 0.00%	6.024ms (±0.52%)	5.982ms (±0.40%)	-0.69%

This comment gets updated every time a new commit comes in.