Comment 9 for bug 668895

su_v (suv-lp) wrote :

> IMHO, removing the <check> in the INX files is a valid workaround.

Just tested with Inkscape 0.47 (osxapp-bundled version): while the presence of a <check> directive doesn't prevent an extension to be listed as available file format, it is ignored nevertheless - i.e. the extension is loaded and executed even if <check> refers to a not-existent script file (e.g. txt2foo.pl).

The <check> directive seems to have been 'broken' earlier. Still - I don't feel comfortable leaving broken code in the sources and removing code that exposes it.

> The last patch was needed because the absolute path test was inverted

To me, the current regression is that in 0.48 (possibly like the underlying cause for the regression in bug #551433), the code does not take into consideration that internally paths are not necessarily absolute - depending on configure options for the build (as defined in 'src/path-prefix.h' [1]) - maybe any tests in the 'src/extension/' code relying on this criteria might have to be reconsidered?

If you prefer, I can file a separate report about the regression in osxapp-enabled builds. I do think however that osxapp-enabled builds only expose issues that have been introduced elsewhere to fix the extension system on win32, and are not specific to the way extensions are executed when called from the bundled, self-contained and relocatable osx application.

[1] <http://bazaar.launchpad.net/~inkscape.dev/inkscape/trunk/view/head:/src/path-prefix.h>