ZipInputStream brokenness

Sun Sep 1 09:47:00 GMT 2002

Adam Megacz wrote:
> But it got vetoed for reasons I don't understand. Apparently people
> would rather have a broken java.util.zip than a working one. Go
> figure.

No, we would rather somebody take half an hour to debug whatever problem
they are having, rather than bitch. I'm assuming it's not an
intermittent problem - how difficult can it be to track it down?
There are no complex hard-to-understand algorthms, except in the
compression part, but I assume the problem is not there.
H.J. Lu used to have a habit of submitting large interconnected patch 
sets for gcc and libg++, without explanation, except to say "it is
needed for Linux". This was incredibly annoying, and we refuse to work
that way. This is is equivalently annoying and non-professional:
"There is a bug; therefore, let us completely replace this
implementation with a differnt one, because the latter doesn't exhibit
*this particular* bug."
Saying an implementation is "broken" because it exhibits a particular
bug is rather naive. Also, it tends to alienate people, which is
not appropriate for a co-operative project. How would *you* like
it if somebody said about your code: "There is an unspecified bug
in your code, so we should replace it."
If you're unwilling to track down the bug, have you at least submitted
a reproducible testcase? Preferably one that can go directly into
the libjava testsuite?
-- 
	--Per Bothner
per@bothner.com http://www.bothner.com/per/