Modernize Code In IDLELib Editor Backspace (and potentially deprecate handling of tabs in IDLE)

[johnzhou721]

In #136061,

cpython/Lib/idlelib/editor.py

Lines 1365 to 1385 in 561212a

	# Ick. It may require *inserting* spaces if we back up over a
	# tab character! This is written to be clear, not fast.
	tabwidth = self.tabwidth
	have = len(chars.expandtabs(tabwidth))
	assert have > 0
	want = ((have - 1) // self.indentwidth) * self.indentwidth
	# Debug prompt is multilined....
	ncharsdeleted = 0
	while True:
	chars = chars[:-1]
	ncharsdeleted = ncharsdeleted + 1
	have = len(chars.expandtabs(tabwidth))
	if have <= want or chars[-1] not in " \t":
	break
	text.undo_block_start()
	text.delete("insert-%dc" % ncharsdeleted, "insert")
	if have < want:
	text.insert("insert", ' ' * (want - have),
	self.user_input_insert_tags)
	text.undo_block_stop()
	return "break"

is mentioned as an area of code with quadratic complexity, which may lead to performance issues; though this is unlikely to be of large impact in practice because attackers with the ability to get a huge file loaded and/or modify configuration files to change the tab width can do much worse things than getting IDLE to hang on hitting backspace.

However, the complexity there comes from a loop which handles situations (files with tabs) that either cannot or should not occur; in the current default IDLE, you can't put tabs, and a space after a tab should just be deleted in a file coming from elsewhere as in indicated in #134874 (comment). Therefore, simplifying the function specification and/or deprecating IDLE to handle tabs (which we should do) will allow elimination of this piece of obsolete code, as indicated in #136061 (comment).

[StanFromIreland]

Can this not just go under #136061?

[johnzhou721]

I am rejecting the refactoring of the handing of space and tab in smart_backspace_event(). Any such project must begin with a separate issue and an detailed analysis of current versus desired behavior, and it must consider the entire block of relevant code. This is lines 1365-1385, which begin with the comment

#134874 (comment) by @terryjreedy

I have not analyzed the current vs desired behaviro yet. Maybe someone else can tackle this. Also maybe make it type-refactor not type-feature.

[StanFromIreland]

type-refactor

deprecating IDLE to handle tabs

I’m sorry but I have never come across a definition of refactoring that includes deprecation?

[johnzhou721]

Yes, but then we can restructure the block of code I linked.

[StanFromIreland]

If a deprecation is required, then it modifies the public interface, which makes it no longer a refactor, as refactoring does not modify user facing functionality.

[johnzhou721]

Sorry for the confusion. Agreed.