On 5/22/16 10:03 AM, jeremyhu@macports.org wrote:
Revision 148902 <https://trac.macports.org/changeset/148902> Author jeremyhu@macports.org Date 2016-05-22 10:03:16 -0700 (Sun, 22 May 2016)
Log Message
inkscape: Revert hack from #51287
If enchant tickles a bug in one of its variants on Mountain Lion, then a workaround should be added to enchant, not inkscape.
I believe there is a mis-understanding here. Even though inkscape uses enchant it requires that the enchant backend be aspell to work correctly and in some code bypasses enchat to access aspell directly. Inkscape checks for aspell and will disable the spell checking functionality if it is not present. This applies to all platforms. Thus the need exists to check for enchant +aspell. The problem with 10.8 is that enchant fails on backend initialization (in the applespell code) if +applespell is enabled. This issue does not occur on 10.9+ but if +aspell is asserted then +applespell is moot because aspell has higher priority in enchant. This is where the check for -applespell comes in. The +applespell issue has not been checked on platforms earlier than 10.8 to my knowledge and so I'm not sure whether this is unique to 10.8. I suggested that the default variants for enchant be +aspell only so that the inkscape default build would succeed in all cases but mcalhoun is resistent to this idea. He believes that +applespell is a more natural default for darwin platforms. Because of this, I believe the variant checks on enchant are valid in inkscape. Hope this clarifies the issue. Dave