Re: [Jilc-dev] Problem

[Gaurav Vaish]

----- Original Message -----
From: "Gopal.V" <address@hidden>
To: <address@hidden>
Sent: Friday, February 22, 2002 13:58
Subject: Re: [Jilc-dev] Problem


: My jdasm code hasn't gone thru any tests at all, all it had to do
: was dump a few code (a Socket program, A SWING program and HelloWorld)
: until now. I didn't give it much thought....

    The course instructor here suggested to use some java-test-units to test the
final compiler -- the first and probably the only thing that comes to my mind is
jUnit. So, writing a test-code using jUnit standards and passing the compiled
code to our compiler to finally yield the il-byte seems to be pretty optimal
strategy to test all the features of java.
    btw, what about the WIDE support? Have do re-looked at it recently?

    yesterday, I was trying to make a table of the opcodes that we will make use
of in IL -- it's no point carrying all the opcodes, quite a few of which we may
never use.

    Also, label management -- I don't think that's going to be an issue. But, we
are ought to implement it.

: Also I find that in the jilc code U gave me, the I,II does
: not print any value ?.

    I have changed it. See the attached file for the bug-fixed release.

: I never tested a class with IINC ;-). Also cut-copy-paste
: from METHOD makes you overlook some simple things....

    I was just thinking of some strategy to proceed for testing.
    I will go through all the parts in detail tonight and/or tomorrow night.

: I'm experimenting with the idea of a 2-pass disassembler lately..
: I'll see how it turns out and back fork it into this code.....

    Two-pass compilers -- hmmm, that's I think the most common technique these
days - at least with most of the compilers that I've gone through recently and
in the past.



Cheers,
Gaurav Vaish
http://home.iitk.ac.in/student/gvaish
http://calendar.yahoo.com/mastergaurav
-----------------------

:
: Gopal.V

jilc-0.0.2.tar.gz
Description: GNU Zip compressed data