slow analysis on big literal array

[staabm]

Bug report

for some reason, we have a source file in the app which contains an array with ~90.000 literal strings.
analyzing this files takes phpstan a big amount of time:

MailAddressList.php
--- consumed 430 MB, total 766 MB, took 194.68 s

Code snippet that reproduces the problem

simplified example, which takes ~1 second to analyze:
https://phpstan.org/r/a1ac48b5-11ac-4f53-b928-2f08bc54affe
it seems my initial sample is a bit too small, because it was cut down by some tool involved.
here is a full script: https://gist.github.com/staabm/ea8b72c50c68648d48820bb7823e98b2

Expected output

faster analysis

[staabm]

looking at a profile suggests, that the path to generalize the literal array is already slow, because too many elements involved to build the generalized key/value types

https://github.com/phpstan/phpstan-src/blob/d37f51dabb5de6d26a5886236e726c0ad6407678/src/Type/Constant/ConstantArrayTypeBuilder.php#L270-L277

https://blackfire.io/profiles/c96451ef-d390-451b-89ed-41189665c8c0/graph

[ondrejmirtes]

Fixed by: phpstan/phpstan-src@0cc87f3

[staabm]

my real world use-case with ~90.000 literal array values still takes 3-4 minutes using the latest dev-1.8. not sure this is worth investigating though.

[github-actions]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.