Context Navigation

Changes between Version 2 and Version 3 of Ticket #26010, comment 7

v2	v3
25	25	> ...returns the earliest error code from the thread's error queue and removes the entry.
26	26
27		I'm a bit confused now, though. Since the documentation appears to be right all along, did the cURL developers misaccuse and blame OpenSSL for their fiasco? Are they now relying on errors returned from the OpenSSL library when SSL/TLS returns an unexpected I/O error? How reliable is this? This is akin to ignoring the original error, since library call errors and I/O call errors are separate and do not depend on one another (although a library call error will probably result in an I/O error, the reverse is not true). I'm most surely missing something by making some assumptions, but how is their strategy of checking a different error stack different from ignoring the read return code with value 0 to begin with?
	27	I'm a bit confused now, though. Since the documentation appears to be right all along, did the cURL developers misaccuse and blame OpenSSL for their fiasco? Why are the developers relying on earlier values of the error queue for a request? Is this reliable?
28	28
29	29	Probably wrong place to ask. Ticket can be closed.