Re: e: Concurrency debugging

Doug Porter wrote:
>  > i have changed the implementation of CRLFInputStream not to use a
>  > pushback buffer, and cleanly separated CRLFInputStream and
>  > LineInputStream functionality (you were right, this was horribly confusing).
> 
> Uh huh. And this is better than the code I had already contributed
> exactly how?

from a brief look at your implementation:

1. CRLFInputStream supports block reads
2. the FSF retains the copyright on the file

the latter is fairly important, if we are going to separate out
functionality into a separate inetutils-java subproject under classpath.
we're still waiting on a decision about this.

don't get me wrong, i don't want to seem ungrateful for your
contributions. what we could really use, though, is some robust test
cases that allow us to locate the bugs you've found. we could also
do with block read support on MessageInputStream if you're still in a
filter-stream writing mood. i don't want to cede FSF copyright on it,
though, even it's found to need a complete rewrite.
-- 
chris burdess