[gentoo-java] jnr-netdb: upstream offers help making ebuilds easier

[Martin von Gagern <Martin.vGagern@gmx.net>]

[-- Attachment #1.1: Type: text/plain, Size: 520 bytes --]

Hi!

As you can see in the attached mail, the jnr-netdb maintainer, Wayne
Meissner, offers his help in making the build process easier. As I only
reported bug https://bugs.gentoo.org/325079 as a user, and haven't dug
deeply into this project, I don't have any strong whishes. But looking
at the ebuild, there is some sed-ing going on, which might perhaps be
avoidable if upstream used variables in a few more places. So maybe one
of you wants to suggest suitable improvements?

Greetings,
 Martin von Gagern

[-- Attachment #1.2: Re: jnr-netdb requires Java 6.eml --]
[-- Type: message/rfc822, Size: 4620 bytes --]

From: Wayne Meissner <wmeissner@gmail.com>
To: Martin von Gagern <Martin.vGagern@gmx.net>
Subject: Re: jnr-netdb requires Java 6
Date: Wed, 23 Jun 2010 08:30:27 +1000
Message-ID: <AANLkTik7bjJNcCoj5hu8caPSBJke3B9agU8isRZGf7Qf@mail.gmail.com>

Thanks for pointing both those issues out!

Yup, the line.length check should be checking for line.length <1 (or
<= 0).  Will apply those changes and update when I get a chance.

btw, if there's anything else needed to make the gentoo build a bit
easier, let me know.

On 23 June 2010 02:07, Martin von Gagern <Martin.vGagern@gmx.net> wrote:
> Dear Wayne,
>
> the jnr-netdb package you develop currently does require Java 6 because
> in jnr/netdb/NetDBIterator.java you make use of String.isEmpty(). It
> would be fairly easy to change that requirement to Java 5 by using a
> String.length() == 0 check instead.
>
> I also noticed that the line in question not only has this compatibility
> problem, but has a bug as well: it contains a check for "line.length <
> 0", with line being an array. The length of an array will never be
> negative, and in case of zero length the check won't prevent subsequent
> access to line[0] in case of a zero-length array, thereby leading to an
> ArrayIndexOutOfBoundsException. Haven't experienced this in real life,
> but you should probably fix the code in any case, and perhaps add a test
> case ensuring proper behaviour.
>
> https://bugs.gentoo.org/show_bug.cgi?id=325079#c1 has a patch addressing
> both these issues. I'll attach a copy here.
>
> Greetings,
>  Martin von Gagern
>


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 262 bytes --]