TYP: fix and improve `numpy.lib._utils_impl` #29181

jorenham · 2025-06-11T17:52:55Z

this resolves some red squigglies in e.g.

Lines 283 to 307 in 44d3cf1

    
           if dtype.names is not None: 
        
               # This is a record array. The .descr is fine.  XXX: parts of the 
        
               # record array with an empty name, like padding bytes, still get 
        
               # fiddled with. This needs to be fixed in the C implementation of 
        
               # dtype(). 
        
               return dtype.descr 
        
           elif not type(dtype)._legacy: 
        
               # this must be a user-defined dtype since numpy does not yet expose any 
        
               # non-legacy dtypes in the public API 
        
               # 
        
               # non-legacy dtypes don't yet have __array_interface__ 
        
               # support. Instead, as a hack, we use pickle to save the array, and lie 
        
               # that the dtype is object. When the array is loaded, the descriptor is 
        
               # unpickled with the array and the object dtype in the header is 
        
               # discarded. 
        
               # 
        
               # a future NEP should define a way to serialize user-defined 
        
               # descriptors and ideally work out the possible security implications 
        
               warnings.warn("Custom dtypes are saved as python objects using the " 
        
                             "pickle protocol. Loading this file requires " 
        
                             "allow_pickle=True to be set.", 
        
                             UserWarning, stacklevel=2) 
        
               return "|O" 
        
           else: 
        
               return dtype.str

charris · 2025-06-11T18:44:40Z

Thanks Joren.

TYP: fix and improve numpy.lib._utils_impl

366b3b3

jorenham added 41 - Static typing numtype Isssue/PR related to numpy/numtype labels Jun 11, 2025

charris merged commit ceb858f into numpy:main Jun 11, 2025
77 checks passed

jorenham deleted the numtype/170 branch June 11, 2025 18:49

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Uh oh!

TYP: fix and improve `numpy.lib._utils_impl` #29181

TYP: fix and improve `numpy.lib._utils_impl` #29181

jorenham commented Jun 11, 2025 •

edited

Loading

Uh oh!

Uh oh!

charris commented Jun 11, 2025

Uh oh!

Uh oh!

	if dtype.names is not None:
	# This is a record array. The .descr is fine. XXX: parts of the
	# record array with an empty name, like padding bytes, still get
	# fiddled with. This needs to be fixed in the C implementation of
	# dtype().
	return dtype.descr
	elif not type(dtype)._legacy:
	# this must be a user-defined dtype since numpy does not yet expose any
	# non-legacy dtypes in the public API
	#
	# non-legacy dtypes don't yet have __array_interface__
	# support. Instead, as a hack, we use pickle to save the array, and lie
	# that the dtype is object. When the array is loaded, the descriptor is
	# unpickled with the array and the object dtype in the header is
	# discarded.
	#
	# a future NEP should define a way to serialize user-defined
	# descriptors and ideally work out the possible security implications
	warnings.warn("Custom dtypes are saved as python objects using the "
	"pickle protocol. Loading this file requires "
	"allow_pickle=True to be set.",
	UserWarning, stacklevel=2)
	return "\|O"
	else:
	return dtype.str

Uh oh!

TYP: fix and improve numpy.lib._utils_impl #29181

TYP: fix and improve numpy.lib._utils_impl #29181

Conversation

jorenham commented Jun 11, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

charris commented Jun 11, 2025

Uh oh!

Uh oh!

TYP: fix and improve `numpy.lib._utils_impl` #29181

TYP: fix and improve `numpy.lib._utils_impl` #29181

jorenham commented Jun 11, 2025 •

edited

Loading