1445553 - More sixgill improvements

Status: RESOLVED FIXED

(Core :: JavaScript: GC, enhancement)

Description

[Steve Fink [:sfink] [:s:]]

Forked off of bug 1443233.

Comment 1

[Steve Fink [:sfink] [:s:]]

Attachment: Cross-check the compile-time plugin headers vs the runtime host gcc

I swear this didn't use to work, but it does now. Make sure we are running sixgill against the same version of gcc it was compiled for.

Comment 2

[Steve Fink [:sfink] [:s:]]

Attachment: Improve lambda check (there can be <lambda(foo)>)

Handle <lambda(foo)>

Comment 3

[Steve Fink [:sfink] [:s:]]

Attachment: File-qualify anonymous structs

I think I had a confusing collision of anonymous structs from different files?

Comment 4

[Steve Fink [:sfink] [:s:]]

Attachment: Hack for lambdas that return their own type

Ok, this is a hack to avoid infinite recursion. There are probably better, more principled ways to fix this, but it doesn't seem worth much trouble.

Comment 5

[Steve Fink [:sfink] [:s:]]

Attachment: Handle the case where DECL_CONTEXT(returnval) is null

I doubt I ever dug enough to figure out what was really going on here, but I was getting a crash.

Comment 6

[Steve Fink [:sfink] [:s:]]

Attachment: wrong header for VOID_CST check, at least for newer gcc

Fix which header we include to check for VOID_CST. With newer gcc, it was getting this wrong.

Comment 7

[Steve Fink [:sfink] [:s:]]

Attachment: Handle a few more compile flags

Bleh. There are a few more -M flag variants in use now.

Comment 8

[Steve Fink [:sfink] [:s:]]

Ok, that was straightforward. I guess I won't do any fixups after all. Thanks!