[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: japi note
From: |
Tom Tromey |
Subject: |
Re: japi note |
Date: |
30 Sep 2004 09:23:24 -0600 |
User-agent: |
Gnus/5.09 (Gnus v5.9.0) Emacs/21.3.50 |
>>>>> "Stuart" == Stuart Ballard <address@hidden> writes:
>> Or even better, could we put up a japitools CVS somewhere hackable?
Stuart> As the author of Japitools, let me add a big "hell yeah" to
Stuart> that one :)
Stuart> Is Savannah accepting new projects yet[1]?
It is.
Stuart> In the meantime, Tom, if you would send me the patch, I'll put
Stuart> out a new tarball. I've always wondered why people were
Stuart> getting bogus results, so I'm very curious to see it :)
This patch is really bogus. It fixes the problem by working around
it. Basically while debugging I found that re-reading the method info
would end up adding an incorrect exceptions attribute to the
MethodInfo. Exactly why, I don't know. This patch works around it by
arranging to only parse bytes once for a given method.
Tom
--- src/jode/bytecode/ClassInfo.java.~1~ 2004-09-30 09:05:56.792464728
-0600
+++ src/jode/bytecode/ClassInfo.java 2004-09-30 09:06:04.665267880 -0600
@@ -286,7 +286,7 @@
/* Since we have to read the whole class anyway, we load all
* info, that we may need later and that does not take much memory.
*/
- howMuch |= HIERARCHY | INNERCLASSES | OUTERCLASSES;
+ howMuch |= FULLINFO;
howMuch &= ~status;
/* header */
if (input.readInt() != 0xcafebabe)
Re: japi note, Stuart Ballard, 2004/09/30