classpath
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: jdiff.sh


From: Stuart Ballard
Subject: Re: jdiff.sh
Date: Thu, 19 Sep 2002 10:06:12 -0400
User-agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.0.0) Gecko/20020615 Debian/1.0.0-3

Giannis Georgalis wrote:
        Hello,
I've written a small shell script that compares the methods,
constructors and attributes (public,protected) of the Java official
API (from the official api specification in html format) to the GNU
classpath corespondent classes (from the source code of GCP).

The script doesn't handle (yet) correctly all the thrown exceptions
that it extracts from the html java documentation, but still the
output is quite usefull.

Wow, that's cool!

Would you consider helping to integrate this capability into japitools (http://rainbow.netreach.net/~sballard/japi/)? Japitools already does full API comparisons but it doesn't have the capability to extract APIs from the documentation - instead, the "japize" tool uses the jode.bytecode library to identify the public members of a given set of packages from the binary jars or zips. But japitools does separate the API-inspection phase ("japize") from the compatibility-testing phase ("japicompat") so it would be easy to offer an alternative to japize that worked on API documentation.

If your script could generate a ".japi.gz" file from API documentation without needing to look at the binaries, that would be a great feature.

I'm in the process of writing a specification for the format of japi files, but this isn't nearly complete enough to base an implementation on yet. The source code for japize is the canonical version until a spec is completed. The release of jdiff means I'll prioritize the spec more highly, so that an interoperable implementation is possible.

I've held off on announcing japitools releases publically because I'm hoping for further testing from a few people who are using it already; I don't want to make a big announcement of something that turns out to be broken, like a few of my recent releases have been :( (Also when I announced it initially, 2 years ago, nobody seemed to be particularly interested). Further testing of the current (0.8.6) release by other groups would help me to make a publically-announced 0.9 release that's stable, correct, and most of all, soon! :)

Stuart.

--
Stuart Ballard, Programmer
NetReach - Internet Solutions
(215) 283-2300, ext. 126
http://www.netreach.com/





reply via email to

[Prev in Thread] Current Thread [Next in Thread]