broken compatibility from jna-3.4.0 to 3.5.0

[Ismael Olea]

Hi:

I've just realized svnkit 1.7.5-v1 doesn't build with jna-3.5.0 while it
worked with 3.4.0.

I'm the only one with this problem?

Here you can find the errors I got now:
http://bsjones.fedorapeople.org/reviews/svnkit/build.log

[Aleksandar Kurtakov]

----- Original Message -----
> From: "Ismael Olea" 
> To: "Discussion list for java related Fedora development" 
> Sent: Tuesday, November 27, 2012 2:04:06 PM
> Subject: [fedora-java] broken compatibility from jna-3.4.0 to 3.5.0
>
> Hi:
>
> I've just realized svnkit 1.7.5-v1 doesn't build with jna-3.5.0 while
> it worked with 3.4.0.
>
> I'm the only one with this problem?
>
> Here you can find the errors I got now:
> http://bsjones.fedorapeople.org/reviews/svnkit/build.log

We have seen such problems many times in many packages. Usually adding a dummy method just to satisfy the signature is enough though in this case it might not be.
Please report it to svnkit developers too.

Alexander Kurtakov
Red Hat Eclipse team
> --
>
> Ismael Olea
>
> http://olea.org/diario/
>
> --
> java-devel mailing list
> [e-mail]
> https://admin.fedoraproject.org/mailman/listinfo/java-devel

[Farkas Levente]

unfortunately from 3.5.0 getFieldOrder() method is required of all
subclass of Structure:-(
you can read more about it here:
https://groups.google.com/forum/?fromgroups=#!topic/jna-users/ju0CyEmiU_4

On 11/27/2012 01:24 PM, Aleksandar Kurtakov wrote:
> ----- Original Message -----
> > From: "Ismael Olea" 
> > To: "Discussion list for java related Fedora development" 
> > Sent: Tuesday, November 27, 2012 2:04:06 PM
> > Subject: [fedora-java] broken compatibility from jna-3.4.0 to 3.5.0
> >
> > Hi:
> >
> > I've just realized svnkit 1.7.5-v1 doesn't build with jna-3.5.0 while
> > it worked with 3.4.0.
> >
> > I'm the only one with this problem?
> >
> > Here you can find the errors I got now:
> > http://bsjones.fedorapeople.org/reviews/svnkit/build.log
>
> We have seen such problems many times in many packages. Usually adding a dummy method just to satisfy the signature is enough though in this case it might not be.
> Please report it to svnkit developers too.
>
> Alexander Kurtakov
> Red Hat Eclipse team
> > --
> >
> > Ismael Olea
> >
> > http://olea.org/diario/
> >
> > --
> > java-devel mailing list
> > [e-mail]
> > https://admin.fedoraproject.org/mailman/listinfo/java-devel
>
> --
> java-devel mailing list
> [e-mail]
> https://admin.fedoraproject.org/mailman/listinfo/java-devel

[Ismael Olea]

On Tue, Nov 27, 2012 at 1:45 PM, Farkas Levente  wrote:
> unfortunately from 3.5.0 getFieldOrder() method is required of all
> subclass of Structure:-(
> you can read more about it here:
> https://groups.google.com/forum/?fromgroups=#!topic/jna-users/ju0CyEmiU_4

If the new release brokes API, should we update instead of keeping on
3.4.0? If so, would be reasonable to keep an aditional jna34 package?

Honestly, I don't have the skills to patch SVNKit sources :-m

[Ismael Olea]

On Tue, Nov 27, 2012 at 1:24 PM, Aleksandar Kurtakov wrote:
> We have seen such problems many times in many packages. Usually adding a
> dummy method just to satisfy the signature is enough though in this case it
> might not be.

I see.
> Please report it to svnkit developers too.

Done: http://issues.tmatesoft.com/issue/SVNKIT-329

[Stanislav Ochotnicky]

Quoting Ismael Olea (2012-11-27 16:24:26)
> > We have seen such problems many times in many packages. Usually adding a
> > dummy method just to satisfy the signature is enough though in this case it
> > might not be.
>
> I see.
> > Please report it to svnkit developers too.
>
> Done: http://issues.tmatesoft.com/issue/SVNKIT-329

Ccing devel so that audience is bigger (I believe this deserves it)

Just for future reference. What you have written in upstream bugreport is
basically: "I have tried to build your software in unsupported way (ant vs
gradle) and used different versions of your dependencies and it failed. I've
done it because other project has some silly rules. Can you fix it?".

Upstream might or might not realize what you are asking for is a simple dependency
bump with API fix (most likely). And they can get annoyed by your request (some
upstreams definitely would). To be precise, this situation is not a bug in
upstream at all, but it is caused by us (Fedora as a distribution). Therefore
it's more of a RFE as far as upstream is concerned. RFEs written as bugreports
can be hard to sell.

This is not just for your benefit. I've seen several such bugreports from Fedora
contributors. They can sometimes make dealing with upstreams difficult, so
please be careful with the wording.