Ambiera ForumDiscussions, Help and Support. |
|
|
|||||
|
Hi, After having tried the app-breaking secureswf and SWF Encrypt, I discovered your project. It's a lot safer and simpler. Though, I've come to many issues: . the parsing doesn't seem complete, class names and package names are mixed. It seems like no type is attached to a variable: I have "destroy" methods in my code, it get replaced, like Papervision's destroy methods. . the ignore list should contain fully qualified class names . there can be only 1 source directory. I would have liked to add the libraries source directories. I can always put all the sources in the same directory. . the packages definitely need to be obfuscated. . I can't use the whitespace removal functionality. It would be great if you could add a semicolon before any "public", "protected", "private" or after each "}" (I don't really know the real solution) If I had to choose (which I know I don't :) ), I'd prefer the last 2 issues fixed. For info, you can look at the doc of Proguard, the reference obfuscator for Java. Look at its command line arguments...its Retrace tool. Your tool is still very usable (the only one actually). Thanks for making it free for 30 days (and then affordable) ! Olivier Allouch |
||||
|
thanks for posting, it is planned already to improve the tool into this direction :) |
||||
|
Any roadmap ? :) |
||||
|
Man, I gave up on your tool (at least for this version) 2 major blocking bugs (without mentioning the annoying renaming of packages): . when a package is skipped, it can contain references to obfuscated variable ! . when a symbol is skipped, the class file itself is still renamed ! I think this kind of bug can be easily fixed. Maybe you should go open source. |
||||
|
[quote=331]. when a package is skipped, it can contain references to obfuscated variable ! . when a symbol is skipped, the class file itself is still renamed ![/quote] Hm, as far as I can tell, this should work, and just tried it myself with a lot of combinations, and it works perfectly. Could you give a short example where this bug occurs? |
||||
|
You can find an example at http://oallouch.free.fr/irrFusca.... Read the Notes.txt file. If you have a problem with the file, send me an email. |
||||
|
Thanks, now I am able to reproduce this, goign to fix it soon. |
||||
|
Update: Ok, this example is quite evil, a lot of problems with case(in)sensitive filenames in windows, local variables with these names of the classes and rules for ignoring the package and one of the class are colliding here, causing this mess. This will not be easy to fix, maybe not fixable at all. The good news is that irrfuscator is planned to be extended soon a lot and will have no problem with this kind of problem in the near future anymore, but this could take some time. As workaround until then just remove the ignore statements (package as well as class), it should work then. |
||||
|
Here's the thing: . I ignore my data sub-package (which I had to rename data2 because it collides with some data packages of Papervision and you don't use fully qualified names) because there's a bug in the treatment of XML XPaths (the last part of the part gets obfuscated) . this data2 package uses classes from the outside, so I have to ignore those classes . and I still have to ignore classes that collide (in a case insensitive way) with package names (in my code and in Papervision) So, I guess if you repair the XML bug (if you want an example...), that should do it :) BTW, if you want a cross platform solution, you should consider coding it in Java (that way I could give you a hand) Olivier Allouch |
||||
|
Sure, an example would be very welcome :) |
||||
|
Actually, I just realized that the bug occurs when an XML tag name collides with the name of an obfuscated class. For example, xmlNode.ns::Item gets Item obfuscated when there's an Item class that's obfuscated. I'll try to avoid ignoring packages by only ignoring those cases or changing some class names. |
||||
|
|
){kind=link}