Incorrect DER output

[pzb]

Incorrect DER output

[sleevi]

@vlm I had initially hoped it'd be more expedient to highlight the bug (from casual reading). When @pzb pointed out it'd been open for over a year and a half, I decided it was worth the hassle to go through our internal open-source licensing process if it would help get it merged :)

[sleevi]

@vlm - This looks like an easy fix, as the SEQUENCE skeleton for uper already appears to handle this logic, since it has to handle the presence bitmap & encoding.

It looks like if

asn1c/skeletons/constr_SEQUENCE.c

Lines 1376 to 1394 in 6b949bb

	/* Fetch the pointer to this member */
	if(elm->flags & ATF_POINTER) {
	memb_ptr2 = (void **)((char *)sptr + elm->memb_offset);
	if(!*memb_ptr2) {
	ASN_DEBUG("Element %s %d not present",
	elm->name, edx);
	if(elm->optional)
	continue;
	/* Mandatory element is missing */
	_ASN_ENCODE_FAILED;
	}
	} else {
	memb_ptr = (void *)((char *)sptr + elm->memb_offset);
	memb_ptr2 = &memb_ptr;
	}
	/* Eliminate default values */
	if(elm->default_value && elm->default_value(0, memb_ptr2) == 1)
	continue;

was uplifted to

asn1c/skeletons/constr_SEQUENCE.c

Lines 530 to 539 in 6b949bb

	if(elm->flags & ATF_POINTER) {
	memb_ptr = *(void **)((char *)sptr + elm->memb_offset);
	if(!memb_ptr) {
	if(elm->optional) continue;
	/* Mandatory element is missing */
	_ASN_ENCODE_FAILED;
	}
	} else {
	memb_ptr = (void *)((char *)sptr + elm->memb_offset);
	}

and to

asn1c/skeletons/constr_SEQUENCE.c

Lines 569 to 574 in 6b949bb

	if(elm->flags & ATF_POINTER) {
	memb_ptr = *(void **)((char *)sptr + elm->memb_offset);
	if(!memb_ptr) continue;
	} else {
	memb_ptr = (void *)((char *)sptr + elm->memb_offset);
	}

this would address @pzb 's issue

(Unfortunately, I cannot contribute that patch myself)

[sleevi]

@vlm Turns out it was messier than I realized - for BOOLEANS with DEFAULT values (ATV_TRUE, ATV_FALSE as literals), no default function was being associated at all. This would also affect cases of PER/XER.

I created #181 to try and fix this, with tests. I'm not terribly thrilled with the tests, but they do highlight the issue and cover both SEQUENCEs and SETs. Could you take a look?

[vlm]

Merged, thank you!

@sleevi if you don't mind me asking, I am curious why couldn't you create a patch yourself and then ended up creating it anyway?