I found a further problem and a workaround to both problems.
The workaround is to use Firefox, which does not suffer from the problem described below. So it seems that a recent Safari update must be the problem. In addition, the ability of a signed applet to use Runtime.getRuntime().exec to launch new browser windows also continues to work in Firefox, but doesn’t work in Safari. So it seems clear that a bug report should go to Apple, not Oracle. Also, there is a good workaround to use Firefox, so we will let our users know that they can stay with Macintosh, but just avoid Safari. If others have input, particularly on ways of coaxing safari into working, that would be appreciated. I wrote: I’ve also demonstrated that the same problem affects a signed Java applet trying to write files on the Macintosh. The write-up is at http://www.segal.org/java/localFileWrite/. Each write-up references the other so as to give a more complete version of the problem. Feedback would be appreciated so as to improve the usefulness of bug reports, as described in the original message. I wrote: A change in MacOS in the past week or two appears to have blocked local file reading by signed Java applets running Java 7. I’ve illustrated the problem with a working example and full source code at http://www.segal.org/java/localFileRead/. The problem does not seem to be related to Java 7 update 45 since it also occurs with Update 40, and that had been working fine until a week or two ago, as assessed by reports from users of a large applet. The demo applet was prepared to demo the problem using concise code. A signed applet is supposed to be able to read and write local files. Although this capability is restricted on some institutional networks, this is the first time these abilities have been restricted for individual users. Does anyone know whether this is a bug, or done purposely by Apple? Are there any workarounds of settings that a user can set to allow such file reading? Since this appears to represent a change by Apple, should this be reported as an Apple bug, an Oracle Java bug, or both?
