John,

I'm building a >100mb assembly and the build is blowing up with an OOME. I upped the limit to 128m and instead of that problem it completely hangs down the track (kill -9). I'm trying again with 256m now.

If I build with 2.1 again, there are no problems. I am able to switch back and forth and get consistent results: dead with 2.2, fine with 2.1. Oddly, though, I had no such problems with 2.2 yesterday.

Did anything change overnight that might leak memory?

- Brett

On 30/03/2007, at 3:50 AM, John Casey wrote:

This is the second attempt, after fixing:

* http://jira.codehaus.org/browse/MASSEMBLY-155
* http://jira.codehaus.org/browse/MASSEMBLY-192
* [file and directory mode parsing]

WRT mode parsing, everything is now parsed using Integer.parseInt ( mode, 8
). If this results in a NumberFormatException, that is wrapped in an
AssemblyFormattingException that indicates that it was a file/dir mode that failed to parse. If the parse succeeds, the resulting mode int is checked
for sanity. That is, cases where group|world have perms that the user
doesn't, or cases where world has perms that the group doesn't (as in,
user:rx, group:rwx, world: rx). This could still break builds where a mode is specified in decimal, but at least there should be some indication of what went wrong...and now they'll know that we're trying to do the right
thing. Also, all of the mode parsing has been consolidated into
TypeConversionUtils, with a TypeConversionUtilsTest to check my work.
Finally, for MASSEMBLY-155, there is a new integration test that verifies
its behavior.


So, let's try this again:

I wanted to call a vote to release a beta version of the new assembly
plugin. There are still some outstanding issues (though not as many as jira would have you believe; they just need tests), but I think we're at around 95-99% backward compatibility and the new features seem to be working well. It's been just sitting in SVN for quite awhile now, and many people are using it directly from there. I'd like to provide better support for those people, and start getting more feedback on exactly what's still broken.

The change list is pretty large, but is mainly concerned with refactoring the plugin away from the old monolithic approach to one that uses phases to handle the different descriptor sections, along with common task classes to
handle behavior shared between phases.

Road Map:

http://jira.codehaus.org/secure/ReleaseNote.jspa? projectId=11126&styleName=Html&version=12617


Tag:

http://svn.apache.org/repos/asf/maven/plugins/tags/maven-assembly- plugin-2.2-beta-1

Staging repository:

http://people.apache.org/~jdcasey/stage/repo<http:// people.apache.org/%7Ejdcasey/stage/repo>

So, let's try 72h +1/+0/-1. Please cast your votes!

Here's my +1.

-john

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to