ExternalAsciiStringResource is gone in upstream V8

[bnoordhuis]

Replaced in v8/v8@299ed09 with 'ExternalOneByteStringResource`. Not sure about the exact version because the V8 people don't tag properly but it was probably a 3.28 release.

This affects iojs because it ships V8 3.30. I'll try to come up with a patch.

[kkoopa]

I'll guess it always was an 8-bit char array, that is, not proper ASCII. If so, a straight replacement should work 1:1, without any filtering or masking out possible high bits. Just need to condition on the right versions somehow. Has io.js got any preprocessor symbol indicating that it is iojs and not joyent node?

[bnoordhuis]

Not at the moment. I'm leaning towards removing the NanNew() overloads for ExternalAsciiStringResource because those won't work across versions anyway. The only other place it's used is in _NanGetExternalParts() and that can probably be made to work through SFINAE.

[kkoopa]

As the type has been removed, it seems like the only way, although I do not like it. It means the 1.5 branch will have to become 2.0, as functionality is removed. The good part there is that All ugly deprecated stuff can be removed, including _NanGetExternalParts, which is only used in the deprecated string converter. Then the changes to NanWeakCallback from #218 have to go in, among with removing the NAN_WEAK_CALLBACK macro.

[bnoordhuis]

@kkoopa I cobbled something together that tries hard not to break the API, please see #223. It probably still warrants a minor bump because I won't vouch all compilers will always infer the type of T in all situations. g++ demonstrably does not so msvc probably does even worse.

[rlidwka]

Any progress on resolving this?

I'm trying to port my binary packages to work under io.js when it's released, and this issue is a blocker. :(

[bnoordhuis]

@rlidwka I updated #223, PTAL.