Improve sprintf return type inference
#3168
Conversation
staabm
force-pushed
the
bug11201
branch
4 times, most recently
from
8dd5fb7 to
a0824ef
Compare
| // XXX should be string because of invalid arguments count | ||
| assertType('non-falsy-string&numeric-string', sprintf('%10$14F', $mixed, $s)); |
There was a problem hiding this comment.
after #3126 is merged we have the necessary PrintfHelper to decide about the invalid number of arguments/placeholders
|
the composer error is valid, as we previously inferred a non-empty-string return type, even though the sprintf pattern is not known at static analysis time. other errors are the same ones are unrelated |
|
This pull request has been marked as ready for review. |
staabm
force-pushed
the
bug11201
branch
4 times, most recently
from
cfc5771 to
fc9baf6
Compare
|
the reported phpstan error is interesstig.. it seems - depending on some unknown factor - reports that we are catching a non-throwable but ignoring this very same error by identifer leads to a "No error with identifier catch.notThrowable is reported on line 200." see the 2 variants at (its unrelated to this PR though) |
|
I think its related to this |
|
There's an opportunity to grab the code from Hoa packages and make a small single modern library with no magic and simple PSR-4 autoloading and replace Hoa in phpstan-src. I might look into it this summer. |
staabm
force-pushed
the
bug11201
branch
2 times, most recently
from
15bbfa6 to
fc9baf6
Compare
ok, what can we do to fix the build until then? |
|
Try to rebase on top of: 3233155 |
looks great, thanks |
There was a problem hiding this comment.
I'm sorry. I see the diff of src/Type/Php/SprintfFunctionDynamicReturnTypeExtension.php and my brain just freezes. I'm not able to review this. Can you do this in smaller steps, or maybe not move that many things around?
"Make the change easy, then make the easy change."
closes phpstan/phpstan#11201
closes phpstan/phpstan#10493