wlarip wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Greetings:
Please feel free to redirect me to another forum if this posting is inappropriate.
You're right on spot here for kaffe's kjc :)
I have been attempting to compile the Kopi-2.1B suite using kaffe's kjc in lieu of jikes. jikes 1.16 seems to have an assertion error based on the absence of some expected byte code generation.
I'd recommend giving jikes 1.19 a go. I've had lots of propblems with jikes releases between 1.13 and 1.18 for building the class library, or with our test suite.
kjc seems to progress well through the suite until it attempts to use jvm initiated jflex-1.4 to generate a scanner entitled KjcScanner.java. In the process of the source generation, it complains of unreachable ASSERT tokens but completes normally. The source is foobar and the compile fails.
Could you try to create some short example that triggers the bug? Our kaffe-extras build of kaffe-1.1.4+ant-1.6.1+patched-kjc-2.1B uses jikes 1.19, afaik, without problems.
I have gotten the impression from list lurking that kaffe's kjc and dms's may have some significant differences. Is this so? if so, what are they?
Yes, kaffe's kjc is dms' kjc 2.1B with a few patches. The patches are available here: http://www.kaffe.org/cgi-bin/viewcvs.cgi/kaffe-extras/patches/ . They are mostly local bugfixes that haven't made it upstream (yet), but may not be as well tested as dms' source code.
cheers, dalibor topic
_______________________________________________ kaffe mailing list [EMAIL PROTECTED] http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
